Home

CMEIAS Ver. 1.28 - The Center for Microbial Ecology

1. 3 a4 Find Objects Ilcon Find Objects A Toolbar Shortcut Select Thresholding Method K Manual Automatic Cancel None image is already thresholded Find Objects Draw a polygon around the region that you wish ImageT ool to search Fig 18 Activate threshold selections to find the image s foreground objects of interest In this example the Search in AOT Automatically select objects and Exclude background features in the Settings gt Preferences gt Find Objects tab page were selected Then the blue polygon arrow was drawn to include all the foreground objects of interest while excluding the bar scale and its associated text 40 5 7 2 Find objects in a binary image If the image is binary has only black and white pixels with grayscale brightness of 0 and 255 respectively select None image is already thresholded in the Find Objects dialog box Fig 18 If you previously selected Search in AOI when setting preferences for Find Objects Settings gt Preferences gt Find Objects gt AOI Options gt Search in AOI see 4 2 1 2 Search in AOD you will be instructed to draw a polygon on the image Use the pencil cursor to draw a thin line polygon left click at every corner that includes all of the foreground objects to be analyzed but excludes the bar scale and associated text optionally plus any other invalid objects present within the image Fig 18 Clo
2. 1 ccsssscccsssseceseeecesaaneeensensesneneseesans 32 4 2 3 5 CHOOSE font aeaoe e T ana i a a E E ea aa a a tacts 33 4 2 3 6 Show object outlines on original iMaQe 11sssccssseecesssecenaneeessseessssanensenssaans 33 A237 oTo e le e E E EE A E E E EOT 33 4 3 Image Settings sisi acne ne cee ec nt vein EE ea E Eaa ne ve e akan Eaa 34 4 3 1 Origin of X Y Cartesian coordinate system in image WINKOWS 0 cscccesceeeeneceneneneneneee 34 4 3 2 Prompt to save untitled iMaQeS 112c sccccsececsnecceeeeceeeeseeeensneeaaeeesseesaneeeaessaneesanees 34 4 3 3 Show full path names in image title Dars 1scccccssceecneneceeansseeenaneeeesanecesaaaeeesaannenes 34 4 3 4 Initial zoom scale factor sonnnaanaaanannnn ev Sic ee ccd vais nnnnnnrnnrennnr Res aes RS 34 IMAGE DISPLAY ADJUSTMENTS CALIBRATION and THRESHOLDING 35 og Feo Tell renege a a SS E PSPE A ee oe 35 5 2 AGIUSt IMAGE Size erronea dea wie vpn aa AE ia ER Ae ANEK Sae aNG 35 5 3 Image TranSlOrmMatiOms icc suede sentenced dv cwccbiuesedsaias det eeddunesnecabtaaneades cdeduonrauaccecemeneens 36 5 4 Adjust Image Contrast Brightness ccceeeee eee eee e eee e seen eee eee eee eeaeeneeeee eee 36 5 5 Negative Image Transformation c cceeeeeeeeeeeeeeee seen eee eeeeeeaeese eee eseesaneeeaeeees 37 5 6 Spatially Calibrate the Image ccecceeeee eee e eee e eee eee ea neat ences eae ena ese eee esse eae 38
3. 1ccccccceccecseeeensnseecnausecessneeeesaaeeesanaeesssenseseesenseneesaes 28 4 2 1 1 Search entire UNA GO isccnaiice seciretenasahica eeuseseevasnassabesuecsdaccadcaeesunendaceacineesaaleres 28 4 2 1 2 Search in AOI Area Of INt r St ssssscssscssssscscccssssssssssesssesssssssessssssaesssessssaees QO 42 2 Search OPtiGNs soicsinceecesives staan e a r O a ENEE EEEE 29 4 2 2 1 Manually select ObjeCts 1ssccccccseccecasseeecnscsecensuscenanaceesaansaaesessssseseneesaaaees 29 4 2 2 2 Automatically select ODjOCtS 1 cccccsscccccsseccesnaceeecssnecesuseeeeseesaaneesssneessaaaeees 29 4 2 2 3 Include objects at edge Of iMAaGe 1 ccsssssecsncseeeeausceessaeessnsneeeesnaneesssneesessanees 29 4 2 2 4 Exclude DaChQroun sci ccesscaiscnssseenscacenavassiepneneacisanaskebadeneusstnnasdecdastenasesipaacneses 29 5 4 2 2 5 Maximum of obj cts esner hiv ete ei each ieee 29 4 2 2 6 Minimum object Size PixlS 12 c 0cccccceeseceeceecna eee seeensneessenssescaaeesanessaeesaeeees 30 4 2 3 Display Opl ON S a cevo seca ia heh ae aaa aed ever a eter eee aaraa a ie Tiaan 31 4 2 3 1 Show object Count in a message DOX ccssseccnnsseceesneseecnaeeeensseeeesaaneeeaanesesnes 31 4 2 3 2 Place object Count in Results WINKOW 111ccccsececcssseseeeaeceeeeeeansseessaneesuanneeees 31 4 2 3 3 Concatenate Object Analysis r SUIts ccccccccssceseecsnseeesnaeeesnansecssanaeeeaanesenses 31 4 2 3 4 Show object numbers on original iMaGe
4. i Histogram Ctrl 5 Line Profile Ctrl L Distance Ctrl D Area Ctrl 4 Angle Ctrl h Object Analysis b Count Black AW hite Pixels Results Miei Count amp Tag Finish by double clicking the final object to be counted 47 6 3 Automatic and Manual Measurement Feature Extraction After finding the foreground objects perform an automatic object analysis to extract selected measurement features on them as follows 1 Select any combination of measurement features 4 1 1 Measurement Feature Preference Page Fig 6 Settings gt Preferences gt Measurement Features tab page that you want CMEIAS ImageTool to extract from the foreground objects found The 7 shape measurement features that must be selected in the object analysis routine used by the CMEIAS morphotype classifier and their check boxes are conveniently enclosed within a gray frame in the measurement feature tab page Fig 6 2 Activate the most recently thresholded image Spatially calibrate the image if measurement features that extract dimensional data e g length width from objects are included in the analysis session so they can be reported in user defined units unnecessary if only shape measurement features are selected since they are dimensionless see 5 6 Spatially Calibrate the Image 3 To perform an Automatic Object Analysis click on Analysis gt Object Analysis gt Object Analysis to extract the selected measurement fea
5. Edit Classification Results tool to reassign the object s classification label using the pseudocolored classification image see 7 2 4 CMEIAS 2 Morphotype Classifier for details When all morphotype reassignments are completed click the Done button to update the classification data in the Results window and then save them where desired XXV APPENDIX Known Problems in Uthscsa ImageTool 1 28 1 The ImageTool 1 28 Results window accurately reports only on the first 498 objects in the image regardless of how many objects are found see 4 1 4 Display of Object Analysis Data in the Results Window 4 2 2 5 Maximum of Objects and 6 2 1 Automatic Object Counting 2 The Results window reports the mean and std dev for all values in a column data array and updates these statistics when the data are concatenated from multiple images see 4 1 4 Display of Object Analysis Data in the Results Window and 6 4 Working with Object Analysis Data in the Results Window These are useful and correct statistics for object analysis data but incorrect for the contingency data reports generated by the two object classifiers since the case rows with the 0 and TOTAL values are included in their computation 3 The pseudocolors used to indicate each object s classification are distinctively optimized for the CMEIAS morphotype classifier but not for the IT CMEIAS 1D object classifier in ver 1 28 The ImageTool version 3 0 upgrade has adopted the CMEIAS o
6. Pause During normal use of CMEIAS you would inspect the pseudocolor coded classification result image of the CMEIAS morphotype classification output and if necessary activate the classification edit feature by selecting Analysis gt Object Analysis Edit Classification Results This will open the Reassign Category Label selection window containing option buttons next to all morphotypes in the CMEIAS morphotype classifier Pause To edit the classification result accurately point the star tip of the magic wand cursor precisely over the pseudocolor coded object you wish to edit and left click it once When properly selected the object will flash on and off in the image to indicate it has been selected useful feature when objects are crowded or small and the option button in the selection box will indicate its currently assigned morphotype In this example coccus will be selected see Fig 9B in the tutorial worksheet Next click the option button corresponding to the new classification assignment for that selected object here select the option button for the Regular Rod which will recolor the selected object as true blue see Fig 9C Finally after all necessary edit reclassifications are made to the pseudocolor coded classification image push the Done button to update the classification data in the Results window Click OK now to open this edit selection box and perform this edit morphotype classification routine
7. The next sections include recommendations to produce good images for quantitative image analysis starting with the sample preparation to immobilize dispersed bacteria on agarose coated slides The goal is to produce phase contrast micrographs containing the refractile bacteria immobilized in the same flat focal plane on the agarose surface and at an ideal recommended spatial density of between 30 170 bacterial cells per field of view If necessary dilute or concentrate the cell suspension using filter sterilized water or culture medium and prepare additional slides to produce this ideal spatial density of bacteria Cell densities lower than 30 per microscope field will require too many images to produce a statistically adequate sampling for the dataset and densities higher than the maximum recommended 170 per field will likely contain many touching cells that must be separated by image editing procedures prior to analysis plus projected images of cells that are likely to be too small lt 30 pixels for morphotype classification 3 2 Preparation of Agarose Coated Slides 1 Wash a high quality agarose e g Boehringer Mannheim LE agarose Cat No 1685651 four times with deionized water aspirate supernatant fluid following sedimentation 2 Prepare a 1 6 w v suspension of the washed agarose in deionized water in screw cap bottles and autoclave them 121 C for 15 min 3 Clean several glass microscope slides with frosted ends and wipe d
8. classification data or Yes if you want to save these object analysis data The latter choice will open a window to designate the name and location of the txt file where the data will be saved Then CMEIAS will perform a morphotype classification of the microbes analyzed in the image and report the classification data in the Results window All options that can be applied to data collected in the Results window are described in section 6 4 Working with Object Analysis Data in the Results Window in the CMEIAS 1 28 User Manual and CMEIAS 1 28 Help file Pause Now click OK to run the Cmeias 2 Morphotype Classifier on the object analyzed image plug in C Program Files Uthscsa ImageTool Plug Ins objclass dll Pause Check the accuracy of the morphotype classification by inspecting the new image containing each microbe pseudocolor coded according to its assigned morphotype The pseudocolor assignments for each of the morphotypes classified by CMEIAS are indicated in the hierarchy outline in Fig 28 of the CMEIAS user manual and Cmeias128help chm file and in Fig 7 in the CMEIAS Tutorial Worksheet If no errors are found after inspecting the pseudocolored classification image you may Print or Save the data in the Results window using the File main menu or Clear Cut or Copy these data to the system clipboard using the Edit main menu If you find an object whose morphotype classification needs correction activate the Analysis Object Analysis
9. gray centroid integrated density min mean median mode max gray level densities and their standard deviation gray centroid x y and centroid x y coordinates New measurement features of CMEIAS Ver 1 28 added to the Object Analysis plug in include maximum curvature length width width length ratio length width ratio area bounding box area ratio eight Fourier descriptors and aspect ratio Fig 6 shows the Measurement Feature tab page with the selected object analysis attributes that must be extracted from each foreground object found within the image to perform a CMEIAS 2 morphotype classification 21 Measurement Features Other Measurement Features TES 0 Aspect Ratio Petinneter T Centroid 2 Y Feret Diameter Gray Centroid amp T Length T Integrated Density width Min Gray Level T Lengthwidth T Mean Gray Level Major Axis Length Median Gray Level Major Avis Angle T Mode Gray Level Minor Avis Length Max Gray Level Measurement Precision decimal places 2 Minor Axis Angle Std Dev Gray Level CHEIAS Morphotype Classifier M Roundness M Elongation IY Compactness 4 I Masinunn Curvature M Width Length IW Area BE Area I Fourier Descriptors Fig 6 Measurement Feature tab page Settings gt Preferences gt Measurement Features to activate the measurement features of CMEIAS ImageTool v 1 28 used in object analysis The specific set of 7 features actually 1
10. right and the micrometer image Z ae bottom at the same magnification a used to produce the 10 um bar scale in Adobe Photoshop that is pasted into 10 um 100X J the binary image of bacteria for spatial calibration Fig 5 shows the use of an image editing program e g Adobe Photoshop to create a black rectangular bar scale of known length e g 10 um is ideal for light microscopical images of bacteria directly on a digital micrograph of a micrometer taken at the same magnification and pixel resolution as the image to be analyzed and then copy paste the bar scale onto an uncrowded corner of the segmented image Note that at high magnification the ends of the bar scale line are made exactly at the same relative position left edge of the slide micrometer vertical lines separated by 10 um Some digital camera software can automatically add a pre calibrated bar scale directly on the digital image Also an Enter Magnification plug in 48KB zip file is available for free download at http www reindeergraphics com index php option com_content amp task view amp id 35 amp Itemid 58 that can be used in Adobe Photoshop version 5 and higher to place scale bars on micrographs Unzip the downloaded file and install it into the plug ins folder of Adobe Photoshop This plugin requires user input information on the image magnification the dots per inch dpi in which the image was acquired and the length of the bar desi
11. to the users specifications 7 analyze CMEIAS data by ecological statistics to quantify the similarities and differences in morphological diversity of microbial communities The tutorial assumes that you are already familiar with the general operation of CMEIAS s host program UTHSCSA ImageTool ver 1 28 while running this CMEIAS training tutorial ImageTool provides an operator manual Help files and object analysis tutorial in its program download to help you learn its operation This training instrument is best utilized when accompanied by two readily available documents These are the CMEIAS 1 28 User Manual Cmeias128 pdf and the CMEIAS 1 28 Tutorial Worksheet CmeiasTutorialWorksheet pdf included in the CMEIAS 1 28 installation The worksheet was made using MS Excel and Adobe Acrobat Before starting this tutorial open the tutorial worksheet adjust its zoom for optimal onscreen viewing and then minimize it to the taskbar on your computer so you can easily access and view its contents while running this tutorial Also it is important to open the ImageTool Results window BEFORE starting this macro so you can view the data as they accumulate Position the Result window worksheet near the right edge of the ImageTool workspace adjust its size to display 3 4 columns and expanded to the full viewable height of your graphical user interface This leaves plenty of room to display and work with the images This tutorial analyzes 3 comm
12. um and the resolution limit for light microscopy is 0 2 um the precision should be set at 1 or 2 decimal places for linear or area measurements respectively Shape measurements are generally reported to 4 decimal places If multiple types of measurements are extracted set the precision to the highest corresponding level 26 before analysis then adjust the precision levels for the other measurement feature columns of data after the image dataset has been copied to the spreadsheet application used in conjunction with CMEIAS ImageTool Image analysis of shape area and linear dimension attributes should be extracted with a precision of 4 decimal places and then their precision should be set to 4 2 and 1 decimal place respectively after being copied to the spreadsheet program Set the precision for the Means and Standard Deviation in the Statistics input field within the Settings gt Preferences gt Precision tab page 4 1 4 Display of object analysis data in the Results Window For object analysis work using CMEIAS ImageTool position the Result window worksheet near the right edge of the ImageTool workspace adjust its width to display 3 4 columns and expand to the full viewable height of your graphical user interface to display the collected data Following object analysis each selected measurement feature appears as a column heading in the ImageTool Results window grid Fig 8 and the corresponding measurement values extracted from
13. 1 28 Training Tutorial Macro Cmeias1 28training Tutorial itm Cmeias128trainingTutorial itm is a recorded macro file designed to operate in UTHSCSA ImageTool ver 1 28 on a personal computer using Windows 2000 XPpro Vista Win 7 32 bit This tutorial exercise provides hands on training to acquaint you with most of the important features offered in CMEIAS v 1 28 used in image analysis and classification of microbes and their ecology You must have administrator rights to run this macro file Beginning CMEIAS users may benefit by running this interactive tutorial a few times to gain the skills needed to work comfortably with CMEIAS The tutorial will describe how to perform the following tasks using CMEIAS v 1 28 1 select the required preference settings 2 perform an automatic object analysis of cell size and shape in images of microbial communities 3 analyze the morphotype diversity of a simple microbial community using the IT CMEIAS 1D classifier 4 optimize the upper class limits of bin ranges to compare the frequencies of cell size distribution in different complex microbial communities using the IT CMEIAS 1D classifier 5 measure the morphotype diversity of complex microbial communities using the CMEIAS 2 morphotype classifier 6 use the CMEIAS pseudocolor coding system to locate rare errors in automated morphotype classification and edit the classification results directly so that the final data outputs fully conform
14. 194 and John Russ s Image Processing Handbook CRC Press Some image processing plugins e g J Russ Image Processing Tool Kit can be installed in Adobe Photoshop or Uthscsa ImageTool and be included in recorded action macro routines to edit images so the objects of interest can be found properly by the threshold segmentation routine Our new CMEIAS Color Segmentation software application is also available for download at the CMEIAS website and described in Gross et al 2009 Microbial Ecology DOI 10 1007 s00248 009 9616 7 printed journal version 2010 Microbial Ecology 59 2 400 414 Its image editing features include a powerful color segmentation routine with wide range adjustment in similarity tolerance brightness threshold adjust hue saturation increase decrease intensity split to RGB HSI YUV chromatic channels add contrast user defined min max object size filter fill holes find smoothen sharpen object edges dilation erosion convert to pseudocolors and emboss The software can edit both 24 bit RGB and 8 bit grayscale images Its graphical user interface and about shield are shown here F CHEIAS Color Segmentation Iof x Eie Edit View Process Filters Window Help lalaya elele eRe al el 2 e Enhle CMEIAS Color Segmentation Version 1 0 Chandan K Reddy Colin A Gross and Frank B Dazzo http cme msu edu cmeias 7 A Copyright c Michigan State University Gy Zoom 1 1 x 5
15. 2003 Diversity of bacterial communities in concentric layers of soil aggregates from conventional till and natural forest ecosystems Ann Mtg Amer Assoc Agronomy Soil Sci Soc Amer Denver CO Gantner S R Schuhegger A Steidle C D rr M Schmid C Langebartels L Eberl F B Dazzo and A Hartmann 2003 In situ production of N acylhomoserine lactones by rhizosphere bacteria and their impact on the bacterial rhizoplane community in tomato roots Structure and Function of Soil Microbiota XXIX 27 28 29 30 31 32 33 34 35 36 37 Gantner S 2003 Microbial ecology of N acylhomoserine lactone producing bacteria in the rhizosphere of tomato plants Ph D thesis Ludwigs Maximilian Universitat Munchen Germany 135 pp Dethlefsen L 2004 Translational power differs between bacteria pursuing different ecological strategies Ph D Doctoral dissertation Department of Microbiology and Molecular Genetics Michigan State University East Lansing Michigan Reddy C K and F B Dazzo 2004 Computer assisted segmentation of bacteria in color micrographs Microscopy and Analysis 18 5 5 7 September 2004 issue Dazzo F B 2004 Applications of quantitative microscopy in studies of plant surface microbiology In A Varma L Abbott D Werner and R Hampp eds Plant Surface Microbiology pp 503 550 Springer Verlag Germany Dazzo F B 2004 Production of anti microbial
16. 51 52 It s handy for beginners to have a color printout of the Fig 28 outline near your computer monitor while learning this CMEIAS Edit Classification routine Once the color gt morphotype recognition linkage is mastered this interactive editing routine becomes efficient and easy to do 61 7 2 5 Editing the type 1 misclassification error in a CMEIAS Morphotype Classification 1 Carefully inspect the morphotype assignments of each object in the pseudocolor coded Classification result image Fig 29D and 30A frequent users of CMEIAS will commit the Fig 28 outline to memory Decide which object s need morphotype class reassignment 2 Select Analysis Main Menu gt Object Analysis gt Edit Classification Results to activate the edit module This will evoke a Reassign Category Label window with features grayed out 3 Activate the pseudocolor coded Classification result image Left click the target object with the tip of the magic wand cursor shown in Fig 30B next to the Other 2 class and below label B The object will blink when selected useful especially when objects are small and or have crowded neighbors and its pseudocolor will indicate its morphotype assigned by the CMEIAS 2 classifier EG Label Morphotype Count Coccus C okei B Spiral C Othe 2 gt Spiral C Curved Rod C Other 3 Curved Rod U shaped Rod C Other 4 D U shaped Rod Regular Rod Other 5 _ Regu
17. Dimensions of the original image at the new reduced print size An image resized this way has adequate pixel resolution for accurate analysis using CMEIAS ImageTool 3 7 Image Editing Before performing image analysis the pixels that define the foreground objects of interest must be found and distinguished from those of background The various image processing steps used to prepare an image so that the foreground objects of interest can be isolated by computer vision is collectively called segmentation Foreground objects in images to be analyzed by CMEIAS ImageTool are ultimately found by a brightness threshold procedure which requires that all of their pixels must have brightness values which lie outside the range that defines image background Grayscale images that satisfy this criterion require no further editing and can be segmented directly in ImageTool by the threshold procedure alone illustrated later in Fig 20 More commonly however images of microbial communities contain pixels of invalid objects or other background noise whose brightness levels fall within the range that defines the foreground objects of interest or the bacterial objects are touching each other thus requiring other image editing steps to achieve segmentation prior to analysis by CMEIAS ImageTool ImageTool provides several image processing routines contrast manipulation sharpening median filter smoothing dilate erode spatial convolution with user defined convo
18. Next define the distance or area to be measured by clicking the mouse once at the origin and intermediate corners and twice at the end for distance of the line or at the final corner to close the polygon for area If the image is spatially calibrated the values reported in the Results window will reflect that calibration Analisi Processing Macro Settings Points Ctrl P Count and Tag Ctrl T Histogram Ctrl H Line Profile Ctrl L Distance Ctrl D Alea Ctrl Angle Chri h ns Object Analysis Count Black hite Pixels Distance Length 57 9 pixels 6 4 Working with Object Analysis Data in the Results Window Before performing object analysis using CMEIAS ImageTool position the Result window worksheet near the right edge of the ImageTool workspace adjust its size to display 3 4 columns and expanded to the full viewable height of your graphical user interface to display the data as they are collected Eight options can be performed on the spreadsheet of object analysis data in the Results window l 2 Manually select edit individual cells using the numerical keyboard followed by Enter Manually select and delete individual rows of data from the grid This is the recommended procedure to delete data from a numbered invalid object that was erroneously included in an object analysis session Edit gt Cut Results of all data in the grid to the system s clipboard so they can be pasted
19. active Image Tool will instruct you to select the objects to be analyzed from those automatically identified by the Find Objects command Although ImageTool will still find all of the objects for you it will only report image analysis data on those foreground objects of interest that you select and will exclude the rest as illustrated in Fig 21 Note in Fig 21C that the annotated number assigned to objects will only include those manually selected 0474336 d 1 2 Select Objects B Imagel ool has found and outlined the objects in the image Click on the objects that you want to analyze Select All Deselect All Cancel Done Fig 21 ImageTool s Manually select objects feature In A only 4 of the objects found 42 51 52 and 55 are manually selected as indicated by being temporarily filled with the magenta color selected for the contour annotation After clicking the Done button B these manually selected objects are reassigned new object numbers C and are registered as such in the Results window D after an object analysis 44 CHAPTER Performing Object Analysis 6 1 Overview of Settings Preferences amp Object Segmentation Before performing automatic object analysis you must specify the various Find Objects preferences Settings gt Preferences gt Find Objects Area of Interest Search Display Minimum Pixel Size specify the Image preferences Settings gt Preferences gt Im
20. and is square in contrast to the smaller rounder silver grains in photographic film Foreground objects in digital images with low pixel sampling density typically have jagged edges hence significantly less accurately defined contours Fig 3 Fig 3 Pixel resolution and sampling density of bacterial objects These images of two bacteria were acquired at low left and high right pixel resolution Note the more jagged edges of the objects sampled at low pixel density Acquisition of microbial images at insufficient pixel resolution for morphotype classification is indicated if the jagged contours of regular rod shaped blue pseudocolored bacteria cause CMEIAS to misclassify them often as yellow pseudocolored prosthecates Avoid this source of error by increasing the image sampling density and pixel resolution of the foreground objects We use 1200 dpi to scan 35mm film and maximum native resolution of digital cameras always keep read only backup copies elsewhere followed by resizing the image in Adobe Photoshop to proportionally increase resolution see Image size and pixel resolution This sampling density creates large image files that require large file storage capacity and computer RAM to save edit and analyze them Images must have adequate magnification and pixel resolution so that even the smallest cell of interest is sampled with sufficient pixel density at least 30 pixels to define its contour for accurate morphotype cl
21. antibodies and their utilization in studies of microbial autecology by immunofluorescence microscopy and in situ CMEIAS image analysis In G Kowalchuk F deBruijn I Head A Akkermans J Elsas eds Molecular Microbial Ecology Manual 2nd Ed Chapter 4 04 pp 911 932 Kluwer Publishers Dordrecht Netherlands Dazzo F B 2004 New CMEIAS image analysis software for computer assisted microscopy of microorganisms and their ecology Microscopy Today 12 3 18 23 Dazzo F B J Liu A Jain A Prabhu C Reddy M Wadekar R Peretz R Bollempalli D Trione E Marshall J Zurdo H Hammoud J Wang M Li D McGarrell J Maya Flores S Gantner C Dowling A B Gomaa and Y Yann 2004 CMEIAS V3 0 upgrade Advanced image analysis software to strengthen microscopy based approaches for understanding microbial ecology 2004 Annual Mtg Long Term Ecological Research in Row Crop Agriculture Michigan State Univ East Lansing MI Dazzo F B 2004 CMEIAS advanced image analysis software to strengthen microscopy based approaches for understanding microbial ecology poster 10th Int Symp Microbial Ecology Cancun Mexico Matsuyama J M Fukuda F Dazzo and S Nakano 2004 Changes in bacterial cell volume and morphological diversity analyzed with an image analysis system in a streambed environment poster 10th Int Symp Microbial Ecology Cancun Mexico Hartmann A S Gantner R Schuhegger A Steidle C D rr M
22. box labeled CMEIAS Morphotype Classifier and deselect all other measurement feature choices Then click Apply and OK Measurement Features CHEIAS Morphotype Classifier Roundness Elongation Compactness Maximum Curvature WiidthLength Area BB Area Founer Descriptors command preferences XIV Pause Next we must instruct ImageTool to perform the CMEIAS 2 object classification In the Settings Preferences Object Classification tab page select Report on CMEIAS Morphotype Classifier using multiple measurement features and Display new image showing objects colored by classification and deselect the other choices on this page It is most important that we include this pseudocolor coded classification result image here since it provides the way to inspect the CMEIAS morphotype classification results and correct any errors if found Click Apply then OK to make these selections Object Classification Attributes to send to results window f Report on classifications using a single measurement feature I Value range classification J Mean value of all objects in class I Number of objects in class i Std dev of all objects in class f Report on CMELAS Morphotyoe Classifier using multiple measurement features T Report on objects IY Display new image showing objects colored by classification command preferences Pause Now let s open the Community A tif image find the foreground obje
23. classifier The type 3 error minimized by proper image editing and thresholding prior to analysis occurs when an interactive brightness threshold routine includes an object as foreground even though it actually is inanimate debris a lysed cell fragment e g ghost cell or an invalid object of background noise that should be excluded from image analysis To make the CMEIAS 2 Morphotype Classifier more flexible and reduce unwanted noise in the final data output a plug in module was implemented in CMEIAS first featured in v 1 27 to permit interactive editing of the classifier results This CMEIAS edit module addresses each of these three major types of classification errors that occur during the automated morphological analysis of microbial communities In this routine the user reviews the morphotype classification of each microbe based on visual inspection of its distinctive pseudocolor coded assignment in the result image Fig 28 and 29DE then activates the edit feature manually selects the misclassified object of interest reassigns it to one of 10 other pre defined morphotype classes by clicking the appropriate option button type 1 error Fig 30 or to a user defined other morphotype class type 2 error details in 7 2 5 Editing the Type 1 Misclassification Error in a CMEIAS Morphotype Classification and 7 2 6 Editing the Type 2 and Type 3 Classification Errors or to an Invalid Object category type 3 error Once the user
24. complex bacterial communities as commonly exists in nutrient enriched habitats containing actively growing bacteria that are larger in size and typically monomorphic did not exist prior to development of CMEIAS This recognition of the need to develop a comprehensive computer aided image analysis system that could extract all the information from images needed to recognize and classify the morphological diversity component of microbial communities came to a pinnacle when I was preparing photomicrographs of the diverse microbial community in the bovine rumen for the cover illustration of the 9 Ed of Bergey s Manual of Determinative Bacteriology Fig 1 Sees ee vies Fig 1 Phase contrast microscopy of bacterial morphotype diversity Actively growing and nutritionally enriched microbial communities contain a large diversity of bacterial morphotypes as shown directly by this phase contrast light photomicrograph of bovine rumen fluid Acquiring an image similar to this one for the cover of Bergey s Manual was my spark of inspiration to develop CMEIAS The challenge was to build a computing tool that could extract all the useful information contained in such community images Witness the vast diversity in this community revealed by direct microscopy That work clearly indicated the following three key points e Contrary to current popular thinking microscopy does reveal significant morphological diversity in complex actively growin
25. dev will not remain current Object analysis data from multiple images can only be concatenated if the same measurement attributes are used throughout Any change in that selection will force ImageTool to ask if you want to save the existing data before it overwrites them with data from the new combination of measurement attributes The reasoning for this action should be obvious once you think about its ramifications 4 2 3 4 Show object numbers on original image If checked ImageTool will place the ordinal number assigned consecutively to each object in the image as the scanning analysis proceeds from the bottom to the top of the image during the thresholding routine see 5 7 Find Objects by Brightness Threshold Segmentation Objects whose bottom most pixels are located along the same horizontal position in the image are numbered consecutively from left to right as illustrated in Fig 12 The annotated number is positioned just above the object centroid s Fig 12 Object numbers on original image If the Untitled Image 3 1 1 Miel Show object numbers on original image 4 preference is selected in the Find Objects tab page ImageTool will assign a consecutive number for 4 a A 2 6 each object found in the image from the bottom up and from left to right along the same horizontal position during the brightness thresholding routine 239 This numbering of objects on the image matches the corresponding object number and associated
26. display a message box indicating the number of objects found in the thresholded image s AOI Fig 19 and 22 ImageTool X G There are 70 objects in the image Resuts RE Number of Objects Fig 22 Automatic object counting in ImageTool The total number of foreground objects in the image is indicated by 1 the top annotated object arrow 2 the message box and 3 the Results Window Third if Settings gt Preferences gt Find Objects gt Place object count in Results window 4 2 3 2 is also selected the mean and standard deviation of object counts for all images in a counting session including the object count of the current image will be displayed in the Results window Fig 22 Since in this example only one image was analyzed the Std Dev is 0 00 If the dataset for counting objects consists of multiple images you can concatenate the object count data for each image by deselecting the Show object count in a message box option in the Find Objects tab page Fig 9 4 2 3 3 Concatenate Object Analysis Results The mean and standard deviation for the dataset will update in the Results window with each new image analyzed see 4 1 4 Display of Object Analysis Data in the Results Window The 4 and 5 ways to obtain the object count automatically are featured in the object analysis and object classification routines described in Section 6 3 Automatic and Manual Measurement Feature Extraction Fig 24 Secti
27. each foreground object found in the image are reported individually in case rows in units that are designated during the calibrate spatial measurement step also see 5 6 Spatially Calibrate the Image The mean and standard deviation for all measured values in each column array are automatically computed with a precision of user defined decimal places specified in the Statistics input field within the Settings gt Preferences Precision tab page and are displayed in the gray filled cells of the first 2 rows above the first object row of data Fig 8 These 2 descriptive statistics are useful for object analysis measurement data but not for morphotype classification frequency count data Fig 8 The ImageTool Results window showing CMEIAS object analysis measurement data extracted from a segmented image of microbes Included are column headings of each measurement feature rows of measurement data for each numbered object and mean standard deviation for each array of measurement attribute data collected The only exceptions to this display design are with the Centroid X Y and Gray Centroid X Y data that each display as 3 columns in the Results window one column of ImageTool design contains both the X and Y coordinate values together delimited by a comma and two adjacent columns of CMEIAS design listing the centroid X and Y coordinates separately Fig 8 This latter CMEIAS output design facilitates the use of these spatial coo
28. for their support and advice 238 Microbial Ecology CMEIAS CMEIAS New generation image analysis software for computer assisted microscopy of microorganisms and their ecology Development team Frank Dazzo Jinhui Liu Bin Yu Olga Glagoleva Anil Jain Dom Trione Jaime Maya Ed Marshall Feng I Liu Chandan Reddy Amar Prabhu Madhu Wadekar Ronen Peretz Jose Zurdo Amit Gore John Urbance Ramesh Bollempalli Donna McGarrell Hass Hammoud Kirsten Kulek Mingfei Li Jun Wang Clay Dowling Lei Gao Nisha Hollingsworth Gang Tang Stephan Gantner Youssef Yanni Guoyu Zhu Colin Gross Elisa Polone Deena Nasr Andrea Squartini Abu Bakr Gomaa Crystal Passmore Chris Monosmith Jen Baric Lei Shan Shin ichi Nakano Chris Meyers Isabel Leader Shiva Zamani Nithin Philips Anwar Baruti Ryan Longueuil Lauren Doherty Savannah Dixon Paul Smith Tianli Du Alessandra Tondello Chris Radek Kevin Klemmer Kylie Farrell Boris Krasnov Jessica McCully Copyright Michigan State University Strengthen microscopy based approaches for understanding microbial ecology at spatial scales relevant to the microbes CHAPTER Introduction amp New Features hh icroscopy is one of the most important techniques in microbial ecology since this is the most direct approach to examine the microbe s world from its own perspective and spatial scale The value of quantitative microscopy in studies of microbial ecology can be increased even furth
29. freshwater systems Ph D Dissertation Instituto Di Ricerca Suelle Acque University of Rome Tor Vergata Imai H K H Chang M Kusaba and ShinIchi Nakano 2008 Temperature dependent dominance of Microcystis Cyanophyceae species M aeruginosa and M wesenbergii J Plankton Research 31 171 178 Mishra R Singh J Jaiswal M Singh YG Yanni and FB Dazzo 2008 Rice rhizobia association Evolution of an alternate niche of beneficial plant bacteria association In I Ahmad J Pitcel and S Hayat Plant Bacteria Interactions Strategies and Techniques to Promote Plant Growth Polone E 2008 Intercellular communication in bacteria nodulating plants of the family Leguminosae Ph D Thesis Dept of Agricultural Biotechnology University of Padu Padua Italy Ruusuvuori P J Seppala T Erkkila A Lehmussola J Puhakka O Yli Harja 2008 Efficient automated method for image based classification of microbial cells IEEE 978 1 4244 2175 Gomes de Costa J 2008 Supervision of transient anaerobic granular sludge process through quantitative image analysis and multivariate statistical techniques Ph D Dissertation University of Minho 231 p Rodrigues JL MA Duffy AJ Tessier D Ebert L Mouton and TM Schmidt 2008 Phylogenetic characterization and prevalence of Spirobacillus cienkowskii a red pigmented spiral shaped bacterial pathogen of freshwater Daphnia species Appl Environ Microbiol 74 1575 1582 Dazzo F an
30. hence the maximum curvature has the minimum angle on the object boundary The angle itself is defined as the angle between two equidistant strings each set at a length of eight pixels emanating from the point In Fig 7a the angle at point D is ZGDH where GD HD To compute the local angle the polygonal representation of the boundary is resampled at a constant interval along the object boundary Length Width Width Length Ratio Length Width Ratio The length of the object should be theoretically computed along its principal skeleton which are the loci of centers of maximal disks contained in the object However in terms of accuracy and computational cost it is not easy to extract a useful skeleton since it is very sensitive to boundary noise The closest approximation to cell length provided by ImageTool is the major axis length defined above also called the longest dimension Because this measurement feature can significantly underestimate the length of curved cells e g line CD in Fig 7a we adopted an alternative adaptive algorithm to measure cell length automatically in CMEIAS This algorithm first classifies the object s shape as being either elongated or rounded based on its roundness value and then applies the appropriate formulas to compute cell lengths and widths for each roundness class In the first step objects are automatically classified into one of two types i elongated if Roundness lt 0 8 or ii round
31. how each object is pseudocolor coded according to its specific CMEIAS morphotype class as detailed in the outline of Fig 7 The classification report displayed in the Results window indicates the frequency count for each microbial morphotype class in this image using text with the same corresponding pseudocolors as illustrated in Fig 7 and the image itself and a useful Total row at the bottom Although the Results window always displays the mean and standard deviation in the CMEIAS morphotype classification they are really not useful in this particular image analysis Pause Next open image Community B tif and perform the same Find Objects Object Analysis and Object Classification tasks open C program files Uthscsa ImageTool Help Community b tif command find objects plug in C Program Files Uthscsa ImageTool Plug Ins objanal dll plug in C Program Files Uthscsa ImageTool Plug Ins objclass dll Pause Witness the awesome computing power of CMEIAS Spend a minute to examine this classification result image and note how CMEIAS assigns a specific pseudocolor to each object according to its morphotype classification Maximize the tutorial worksheet and scroll up to revisit the hierarchical classification scheme in Fig 7 again to relate these pseudocolor coded assignments to the other characteristics that distinguish each morphotype XVI Pause Extensive testing indicates that CMEIAS classifies microbial morphotypes in properl
32. in microbial communities since only rarely would a community contain more than 5 bacterial morphotypes not recognized by CMEIAS at any one time Also since the 11 morphotypes included in the CMEIAS classification scheme include all those that are common plus most that are uncommon equal to the morphotypes represented by 97 of all prokaryotic genera in Bergey s 9 Ed Manual of Determinative Bacteriology only on rare occasions will the type 2 error occur in microbial community analysis It has occurred only once in several years of community analysis by developers of CMEIAS Furthermore the rarity and unique character of other bacterial morphotypes make them easy to find These characteristics justify the interactive manual design of this CMEIAS classification edit feature Also use this edit feature to classify rare morphotypes of eukaryotic microbes when they are included in the community analysis 63 Eliminate the type 2 error in morphotype classification results as follows 1 Follow steps 1 4 in to activate the edit feature module and select the object whose morphotype matches none of the 11 morphotypes classified automatically by CMEIAS 2 Reassign the object and all others of the same unique morphotype to class L Other 1 3 Repeat steps 1 2 for any other object s with a unique morphotype s to class M Other 2 and then class N Other 3 etc Up to 5 other unique morphotypes can be added to the classification result
33. object analysis data reported in each row of the Object Analysis Results window and is also useful for interpreting the results of further analysis functions allowing you to visually match object numbers with corresponding object analysis data Note the numbers are not actually part of the image itself but are overlay annotations that remain only while the image is opened They will not appear on a direct printout of the image File gt Print Image on an image saved directly in ImageTool File gt Save Image As or when the image is copied Edit gt Copy Image to another program that can accept it To capture these colored annotations plus the object outlines described in 4 2 3 6 Show Object Outlines on Original Image in the displayed image click the Print Screen keyboard key to copy the entire monitor display to the system s clipboard open an image editing program e g Adobe Photoshop create a New blank image File gt New default will be 72 dpi edit gt paste the image and then flatten and crop the area of interest from within the new image Turn this Show s feature on for CMEIAS image analysis 4 2 3 5 Choose font Double click this command button to display a dialog box allowing you to select the font type style color and size of the annotated object numbers 14 pt Tahoma blue regular displays well 4 2 3 6 Show object outlines on original image This ImageTool feature introduces a thin outline of
34. object count in a message box When checked ImageTool will display a message box that reports the number of objects found by the threshold operation Select this feature only when needed since it adds an interactive step you must click the message box OK button to complete each image analysis cycle This feature helps to keep track of the cumulative number of objects analyzed and to avoid the 498 object limit problem when concatenating object analysis data in the Results window see 4 2 2 5 Maximum of Objects 4 2 3 2 Place object count in Results window Check this option only if you want to collect object count data in the Results window The means std dev of object counts per image are updated in the Results window after each image is analyzed in the same work session Note only one type of data can be displayed in the Results window at a time Thus if the Results window already contains measurement or classification data from a previous analysis a dialog box will display asking if you want to save these previous data before displaying the object count data A yes response will open a second dialog box to enter the name and location of the data txt file to be saved Regardless of your answer to this question all the measurement or classification data in the Results window from the analysis of the previous image will be automatically disposed and overwritten with the new object count data when the Place object count in Results
35. or Clear Cut or Copy these data to the Windows clipboard using the Edit main menu XXII APPENDIX IV CMEIAS 2 Morphotype Classification Macro CmeiasMorphotypeClassification itm If this is the first time you are using CMEIAS we recommend that you perform the exercises in the CMEIAS 1 28 Training Tutorial Macro Appendix I beforehand using the images provided in the program download and installation You must have administrator rights to run this macro file If you want to view the data while they are being extracted from objects in your image optional recommended then before starting this macro you should maximize the Results window position its worksheet near the right edge of the ImageTool workspace and adjust its size to display 3 4 columns and expanded to the full viewable height of your graphical user interface Pause This macro can help guide you through the steps to perform a morphotype classification on objects in your own images using the CMEIAS morphotype classifier It is a supervised hierarchical tree classifier that categorizes microbes in the image according to their complex morphology based on pattern recognition algorithms that use 14 different shape measurement attributes featured in CMEIAS ImageTool v1 28 The classification output data consist of object frequency counts per morphotype class and a pseudocolor coded classification result image that can be used to inspect the classification result and edi
36. own images First open the first image you wish to analyze command open Pause In the Find Objects tab select the parameters you want ImageTool to use to find your objects of interest Remember to select Search in AOI and Automatically Select Objects if you wish to analyze objects in an image that also contains a bar scale for spatial calibration command preferences Pause In the Measurement Features tab select the attribute s you d like to extract from each object found in the image Also indicate the precision decimal places to report your data command preferences XIX Pause Now perform a spatial calibration of your image if you are analyzing objects using a size attribute e g um for cell length If analyzing by a shape or grayscale level attribute e g width length mode gray level it is not necessary to spatially calibrate the image If the latter is the case click the Escape key on your keyboard to bypass this spatial calibration step or draw a line of any length on the image select pixels and accept its default dimension command spatial calibrate Pause Now select the thresholding method e g Manual or None image is already thresholded and perform the brightness threshold procedure to find your foreground objects of interest command find objects Pause You are now ready to perform the object analysis on the thresholded image and view the data in the Results window You may Print or Sa
37. text and instructions for the core files of UTHSCSA ImageTool this updated user manual a self executable Cmeias128help chm file numerous calibration files for use with the IT CMEIAS 1 classifier several revisions to the training tutorial and other macros that improve their usefulness on an international scale and the Cmeias128setup exe wizard to simplify the installation of the CMEIAS v 1 28 upgrade of UTHSCSA ImageTool plus all the other new features of CMEIAS v1 28 described above Also CMEIAS Color Segmentation is now available to segment foreground objects in complex images Feel free to send feedback on CMEIAS to Frank Dazzo at the CMEIAS website at lt cmeiasfd msu edu gt so we can consider it in our upgrades currently under development CHAPTER Requirements Download and Installation 2 1 System Requirements Hardware and Windows Operating System The minimum requirements to operate CMEIAS v 1 28 image analysis within the Cmeias v 1 28 upgrade of the host program UTHSCSA ImageTool ver 1 28 include e a PC with 32 bit operating system Windows NT 4 0 Service Pack 6A 2000 XPpro Vista 7 at least 256 Mb RAM e a monitor displaying 256 colors or higher with at least 800 x 600 pixel resolution ImageTool supports various Window s compatible printer and twain compliant input devices This user manual is written with instructions for CMEIAS ImageTool v 1 28 operating in Windows 2000 and so end users must adjust ins
38. text data file to be saved 9 After entering your preferred response the contingency table of classification data in the Results window will update reflecting the change s you made in the edit routine in Fig 30D the frequency counts for cocci and regular rods are adjusted by one unit 10 Activate the Results window and select Edit gt Copy Results to copy the data to the system s clipboard so it can be pasted into a spreadsheet application for storage and analysis 11 Optional You can also copy the classification result image Edit gt Copy Image or click the keyboard PrintScreen and paste it into an imaging application that can accept it e g Adobe Photoshop Once cropped and selected it can then be saved e g jpg or gif file and or copied directly into the spreadsheet workbook containing the classification data for future reference 7 2 6 Editing the type 2 unrecognized class and type 3 invalid object classification errors Accommodation of the type 2 error has the greatest impact in expanding the range of microbial communities that can be analyzed by CMEIAS This was accomplished by adding five other user defined classes of microbial morphology labeled as L through P corresponding to Other 1 through Other 5 in the CMEIAS edit module interface and assigning a unique pseudocolor to each of them Fig 29C We consider this level of flexibility to be sufficient to handle most bacteria
39. this feature can efficiently eliminate small invalid objects and image background noise during the brightness threshold routine Optimize this setting by first performing a manual Area analysis Analysis gt Area of the smallest foreground object of interest in the image using the default units of pixels see 5 6 Spatially Calibrate the Image and 5 8 Manually Select Objects and then by specifying a slightly lower value in this Minimum Object Size input field so all smaller objects representing image noise are automatically excluded Note both minimum and maximum size of object filters are featured in CMEIAS Color Segmentation At least 30 pixels per object Fig 9 are needed to provide sufficient pixel sampling density for CMEIAS to accurately define the object s contour and shape for morphotype classification see 3 1 Image Requirements for Image Analysis Objects containing less than 30 pixels commonly have sharp jagged edges in this case pixels are large relative to the object itself often causing CMEIAS to misclassify certain morphotypes e g regular rods will commonly misclassify as prosthecates The setting of minimum size can be lt 30 if objects are only counted and analyzed but not classified morphologically If necessary follow the procedure described in 3 6 Image Size and Pixel Resolution to increase the image s pixel resolution before image analysis and morphotype classification 30 4 2 3 Display options 4 2 3 1 Show
40. threshold procedure in ImageTool This image editing step must precede CMEIAS morphotype classification ImageTool Ver 1 28 features some image editing routines but digital images of microbial communities in environmental samples typically require additional interactive editing using other image processing programs e g CMEIAS Color Segmentation Adobe Photoshop etc to achieve full segmentation e g splitting of touching objects removal of invalid objects adjusting the background pixels to brightness values that lie outside the range that defines the foreground objects of interest Consult John Russ s Image Processing Handbook 2002 4 ed for the theory and practice of digital image segmentation 4 Before performing object analysis and object classification in ImageTool CMEIAS the user must select various setting preferences Settings gt Preferences gt on tab pages labeled Find Objects Image Measurement Features and Object Classification then open and spatially calibrate the image and find the objects of interest by the brightness threshold segmentation step s After extraction and computation of selected measurement attributes from each foreground object the quantitative image analysis data reported in the ImageTool Results Window grid are ready to copy to the system s clipboard and paste into a Windows compatible spreadsheet program where they can be processed further and analyzed statistically
41. use of this feature to transform the original positive grayscale immunofluorescence micrograph that contains fluorescent bright bacteria against a dark background into the corresponding negative image containing dark bacteria against a bright background The ImageTool brightness threshold procedure see 5 7 Find Objects by Brightness Threshold Segmentation can be applied to either image to find the foreground bacterial objects of interest for object analysis or object classification Fig 16 Negative Image Transformation Transformation of the positive grayscale image left panel positive immunofluorescence to the inverted negative grayscale image right panel 37 5 6 Spatially Calibrate the Image How Long Is The Line Length fio Units C Manometers Inches Micrometers Feet C Milimeters Yards C Centimeters Miles Pixels Meters C Light Years C Kilometers Astronomical Units Cancel Draw a line of known length Fig 17 Steps to spatially calibrate the image in ImageTool before it is thresholded and analyzed The default measurement unit for ImageTool is the pixel Use the Settings gt Calibrate Spatial Measurements feature to spatially calibrate the image i e automatically convert image pixel dimensions to the selected unit for all measurements having dimension This calibration is valid only for the image on which it is performed Follow these steps to spatially c
42. 28 upgraded files User developed files e g user developed itm macros or ocd calibration files are saved in a new folder named Backup within the ImageTool directory The CMEIAS dll plug in files remain unchanged All Programs SU ate eC aC cere EE CHEIAS Help fig CMEIAS IT 1 28 TG Tutorial Worksheet CMEIAS IT 1 28 Uninstall TG User Manual amp visit website Fig 2 Desktop shortcut icon and start menu items available when CMEIAS ImageTool is installed using the Cmeias128setup exe program A quick way to verify that the CMEIAS v 1 28 upgrade of ImageTool has been installed properly is to display its About shield Help gt About ImageTool which should indicate that it is the ver 1 28 CMEIAS upgrade Also the unique CMEIAS plugin options for object analysis and object classification should display in the corresponding tab pages Main Menu Settings gt Preferences gt Measurement Features or Preferences gt Object Classification as illustrated in Figs 6 and 25 respectively CHAPTER Microbial Sample Preparation Microscopy and Image Preparation Semi automatic image analysis of microbes can be principally divided into five stages image acquisition and digitization Interactive image editing and segmentation to locate the foreground cells of interest automatic extraction of selected measurement attributes from each object found object analysis Classification of different cell types object cla
43. 34y 86 RGB 13 55 14 Pixels Sampled 0 Time s 0 130 560 450x24 WY 3 8 Adding a Calibrated Bar Scale Images intended only for shape analysis morphotype classification need no spatial calibration default unit is the pixel since shape measurements are dimensionless and classification data are of the frequency count type On the other hand images should be spatially calibrated when measuring dimensional attributes so the data reported will be automatically converted from the default dimension of pixels into the user defined units of the measurement feature used no matter what zoom factor is chosen for the image analysis This can be done in ImageTool by either of two ways 1 spatially calibrate the image at high zoom using a magnification bar scale of known length embedded directly in it or 2 load a spatial calibration file saved from a previous image of a slide stage micrometer acquired at the same magnification and pixel resolution Settings gt Load Spatial Calibration gt select appropriate itc file gt Open Use of a spatial calibration file when analyzing many images with the same size magnification and pixel resolution eliminates the need to add a bar scale and calibrate each image Fig 5 Adding a calibrated bar scale to an image This figure shows a phase contrast grayscale image of a methanogenic bioreactor community left the corresponding edited image n with segmented foreground objects aN
44. 4 Take note of a descriptor for each Other X morphotype reassignment 5 When all morphotype classification editing is complete on the classification result image click Done and copy the updated data in the Results window to your spreadsheet application 6 In the same worksheet of your spreadsheet application click the column heading labeled Other 1 etc and rename the morphotype appropriately To illustrate the type 2 error we evaluated 17 valid but rare bacterial shapes that are not included in the CMEIAS Morphotype Classifier Liu et al 2001 Accommodation of the first 16 bacterial shapes labeled A P in community images only required the second editing scenario described above Recognition of the pseudocolor coded assignments directly indicates that CMEIAS misclassified them as follows A as an ellipsoid B and P as unbranched filaments C as a prosthecate D as a Spiral E and F as cocci G as a curved rod H as a regular rod I L as clubs and M O as branched filaments The reasons for most of these assignments are indicated in the hierarchy outline of characteristics for classification of the CMEIAS morphotypes see No software is perfect inner contour outer contour Fig 31 Seventeen examples of rare microbial morphotypes not supported by the CMEIAS Morphotype Classifier A square or cube B almost completely enclosed ring C dumbell D sigmoid E flattened spheroid F lobed spheroid G sickle H lanceolat
45. 4 eight Fourier Descriptors are grouped together as one selection enclosed within the gray line frame on the left portion of the Measurement Features tab page represent those required to operate the CMEIAS 2 Morphotype Classifier Measurement attributes selected on this tab page are displayed as individual column headings and the corresponding data extracted from each object found within the image are displayed in case rows in the object analysis Results Window see 4 1 4 Display of Object Analysis Data in the Results Window 4 1 2 Definitions of measurement features for object analysis The measurement features that are important for image analysis of microbial morphotypes are Area Area of the object measured as the number of pixels scaled to the user defined unit for image calibration in the polygonal approximation of the cell This measurement of size tends to slightly over estimate the object s true area because the borders of the pixels may extend beyond the true perimeter of the cell Perimeter Length of the outside contour of the object represented as a polygon in the digital image Roundness also called circularity or shape factor Computed as 47 Area Perimeter This shape feature measures the degree of object roundness Values lie between 0 and 1 The greater the value the rounder is the object a2 Major Axis Length The maximum distance between points on the object s boundary corresponding to vector CD in Fi
46. 5 7 Find Objects by Brightness Threshold Segmentation cceeeeeeeeeeeees 40 5 7 1 Activate threshold SClCCtiONS 1scccccnceccecanneeessaceesnausecesansseesaaeeessnsnetseaaaaaaeesans 40 5 7 2 Find objects in a binary Image ies desieice tenses is sav haw stks Sec piste este eens ee eel 41 5 7 3 Find objects in a NON DINALY fMAGC 111 cccecceccecaneceennaceesnanecessneeeessaneeensnessanessnaneeees 42 5 8 Manually Select ODjCCts csais cdi se Ascencesee a edee iach sens eta desusdeten eae uacuasaeceeds 44 PERFORMING OBJECT ANALYSIS 0 ccccceeeeeeeeeeeeeeeeeeeeeeeeeeeenaneaeeeeenaenaeenaes 45 6 1 Overview of Settings Preferences amp Object Segmentation sseeeeeeee 45 6 2 Object COUT isiin enana kareni ence and daceadeckdddapeiestacttenteweononestwuchecuantaue 45 6 2 1 Automatic Object Counting sss0nnsa11nrennnnonnnnnnrrennnn een nnnrennnnnrnnrrnnnnnnnnnnnn nennen eenean 45 62 2 Manual Object Counting n nirien a eae ceed 47 6 3 Automatic and Manual Measurement Feature Extraction ccccceseeeeeeeeeeeees 48 6 4 Working with Object Analysis Data in the Results Window eeeeeeee 50 7 PERFORMING OBJECT CLASSIFICATION 0 cceceseeeeeeeseeeeeeeeeeeeesneeeeeeeeeeeeeeens 51 7 1 ImageTool CMEIAS 1D Object Classifier cccceccsseeeeeeeeeeeeeeeeeeeeeeeeeaaeaaaes 52 7 1 1 Using the IT CMEIAS 1D Object Classifier for cell size classifica
47. A Squartini V Corich A Giacomini F De Bruijn J Rademaker J Maya Flores P Ostrom M Vega Hernandez R I Hollingsworth E Martinez Molina P Mateos E Velazquez J Wopereis E Triplett M Umali Garcia J A Anarna B G XXVIII 17 18 19 20 21 22 23 24 25 26 Rolfe J K Ladha J Hill R Mujoo P K Ng and F B Dazzo 2001 The beneficial plant growth promoting association of Rhizobium leguminosarum bv trifolii with rice roots Austr J Plant Physiol now Functional Plant Biology 28 845 870 T bler T M Schl ter O Dirsch H Sievert I B amp ouml senberg E Grube J Waigand J Schofer 2001 Balloon protected carotid artery stenting Relationship of periprocedural neurological complications with the size of particulate debris Circulation 104 2791 2796 McDermott T R and F B Dazzo 2002 Use of fluorescent antibodies for studying the ecology of soil and plant associated microbes In C Hurst R C Crawford G R Knudsen M J McInerney and L D Stetzenbach eds Manual of Environmental Microbiology 2nd ed Chapter 28 p 615 626 American Society for Microbiology Press Washington DC Reddy C J Liu M Wadekar A Prabhu D Trione E Marshall J Zurdo F I Liu J Urbance and F B Dazzo 2002 New features of CMEIAS innovative software for computer assisted microscopy of microorganisms and their ecology 2002 Annual Mtg Long Term Ecologic
48. CMEIAS Ver 1 28 User Manual Custom Plug ins operating in UTHSCSA ImageTool Ver 1 28 Advanced Image Analysis Software Designed to Strengthen Microscopy Based Approaches for Understanding Microbial Ecology F B Dazzo D Trione E Marshall amp J Zurdo Center for Microbial Ecology Michigan State University East Lansing MI 48824 USA CMEIAS is copyrighted by Michigan State University All rights reserved h C fi The MSU Center for MTOE An NSF Science and Technology Center E WN V E R S T Y Table of Contents CMEIAS License Agreement Michigan State University c0csscccseeeeeeeeseeeeeeess 5 UTHSCSA ImageTool License Agreement ccssssssssssseeeeeeccsreceeeeeeeseeeeeneceeeneess 6 Background Reading Acknowledgments About CMEIAS Shield scsseeeeeeeeeees 7 1 INTRODUCTION amp NEW FEATURE So issciccicncsvatinvacnciucwexeeasanoucuncontensbecsecoutenvowsenadannn 8 2 REQUIREMENTS DOWNLOAD and INSTALLATION csceeeeeeeeeeeeeeeeeeeeeeeeeeee 11 2 1 System Requirements icicssscsisecssescstacsietiensiddavenssendstavssccnsnsestetbend vei edessendvedusanssiien 11 2 2 Download amp Installation of ImageTool CMEIAS ver 1 28 cccceeeeeeeeeeeeeeeees 12 3 MICROBIAL SAMPLE PREPARATION MICROSCOPY amp IMAGE PREPARATION 13 3 1 Image Requirements for Image AnalySis 0 ecceeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee eee 13 3 2 Preparation of Agarose Coate
49. CMEIAS morphotype classifier using multiple measurement features and Display new image showing objects colored by classification 2 Open the 8 bit grayscale image select Analysis gt Object Analysis gt Find Objects and threshold 5 7 Find Objects by Brightness Threshold Segmentation to find all foreground microbes of interest in the image There are 170 microbes in the example image Community A in Fig 29 3 Select Analysis gt Object Analysis gt Object Analysis to perform the object analysis on the image Within a few seconds the values of the various shape attributes will be extracted from each numbered object found and displayed in the Results window 4 Make the same image active and then perform the object classification Analysis gt Object Analysis gt Object Classification This should take only a second or less of computing time am Results iC Label Morpho Mean Std Dev D U shaped Rod 5j _ Untitled Image 1 1 4 IEIES Fig 29A C A thresholded binary composite image B reconstructed classification result image with each foreground object pseudocolor coded according to its morphotype class assignment against a black background C classification Result window displaying each class label corresponding morphotype name object count in each class and total number of objects Figs 29D and 29E show the 16 pseudocolors used in the CMEIAS 2 Morphotype Classification 58 Rud
50. Community 6 8 March 2000 Univ of Tokyo Japan 3 Fernandez A S Hashsham S Dollhopf L Raskin O Glagoleva F B Dazzo R Hickey C Criddle and J M Tiedje 2000 Flexible community structure correlates with stable community function in methanogenic bioreactor communities perturbed by glucose Appl Environ Microbiol 66 4058 4067 4 Hashsham S A Fernandez S Dollhopf F B Dazzo J M Tiedje R Hickey and C S Criddle 2000 Parallel processing of substrate correlates with greater functional stability in methanogenic bioreactor communities perturbed by glucose Appl Environ Microbiol 66 4050 4057 Hashsham S T Marsh S Dollhopf A Fernandez F Dazzo R Hickey C Criddle and J Tiedje 2000 Relating function and community structure of complex microbial systems using neural networks Int l Symp on Establishment and Evaluations of Advanced Water Treatment Technology Systems Using Functions of Complex Microbial Community 6 8 March 2000 Univ of Tokyo Japan 6 Dazzo F and J Wopereis 2000 Unraveling the infection process in the Rhizobium legume symbiosis by microscopy In E Triplett ed Prokaryotic nitrogen fixation a model system for the analysis of a biological process Chap 19 pp 295 347 Horizon Scientific Press UK XXVII 10 11 12 13 14 15 16 Dazzo F B Y G Yanni R Rizk F De Bruijn J Rademaker A Squartini V Corich P Mateos E Martinez
51. K Kulek L Gao R Bollempalli D McGarrell Y Yanni A Squartini E Polone S Gantner A Smucker S Nakano 2007 CMEIAS 3 0 Advanced computational tools of image analysis software designed to strengthen microscopy based approaches for understanding microbial ecology 2007 Long Term Ecological Research All Scientists Meeting May 14 Kellogg Biological Station Hickory Corners Michigan Amalfitano S S Fazi A Zoppini A B Caracciolo P Grenni and A Puddu 2007 Responses of benthic bacteria to experimental drying in sediments from Mediterranean temporary rivers Microbial Ecology DOI 10 1007 s00248 007 9274 6 Online First Gioacchini P L Manici W Ramieri C Marzadori and C Ciavatta 2007 Nitrogen dynamics and microbial response in soil amended with either olive pulp or its by products after bio gas production Biology and Fertility of Soils 43 621 630 Dazzo F B 2007 Visualization of the rhizoplane microflora by computer assisted microscopy spatial analysis by CMEIAS image analysis In P Schwinger and R Finlay eds Chapter 4 1 XXXII 61 62 63 64 65 66 67 68 69 70 71 Microbial growth and visualization of bacteria and fungi COST 631 Handbook of Methods Used in Rhizosphere Research Section 41 Swiss Federal Research Institute Birmensdorf Switzlerland pp 382 383 Amalfitano S 2007 Structure and function of benthic microbial community in highly variable
52. Molina E Velazquez J Biswas R Hernandez J K Ladha J Hill J Weinman B Rolfe M Vega Hernandez J J Bradford R I Hollingsworth P Ostrom E Marshall T Jain G Orgambide S Philip Hollingsworth E Triplett K Malik J Maya Flores A Hartmann M Umali Garcia and M L Izaguirre Mayoral 2000 Progress in multi national collaborative studies on the beneficial association between Rhizobium leguminosarum by trifolii and rice In J K Ladha and P M Reddy eds The Quest for Nitrogen Fixation in Rice International Rice Research Institute Manila The Philippines pp 167 189 Liu J F I Liu E Marshall and F B Dazzo 2000 CMEIAS Software for Computer Assisted Microscopy of Microbial Communities Annual Mtg Long Term Ecological Research in Row Crop Agriculture Michigan State Univ East Lansing MI Wopereis J E Pajuelo F B Dazzo Q Jiang P M Gresshoff F J De Bruijn J Stouggard and K Szczyglowski 2000 Short root mutant of Lotus japonicus with a dramatically altered symbiotic phenotype Plant J 23 97 114 Karcher D E 2000 Investigating causes and cures for localized dry spots on creeping bentgrass putting greens In Investigations on Statistical Analysis of Turfgrass Rating Data Localized Dry Spots of Greens and Nitrogen Application Techniques for Turf Ph D Dissertation Dept of Crop amp Soil Sciences Michigan State University East Lansing Michigan Liu J F B Dazzo O Gla
53. Schmid C Langebartels F B Dazzo and L Eberl 2004 N Acyl homoserine lactones of rhizosphere bacteria trigger systemic resistance in tomato plants In Biology of Molecular Plant Microbe Interaction Vol 4 pp 554 556 Eds B Lugtenberg I Tikhonovich N Provorov IS MPMI St Paul Minnesota USA Janssens F and F B Dazzo 2004 Image Analysis Morphometry and Classification of Scanned Collembola Samples Applied to Specimen Identification http www geocities com fransjanssens projects imagproc htm 38 39 40 41 42 43 44 45 46 47 48 49 Dazzo F B 2004 CMEIAS advanced image analysis software to strengthen microscopy based approaches for understanding rhizoplane microbial ecology Rhizosphere 2004 Perspectives amp Challenges abstract Munich Germany Gantner S M Schmid C D rr R Schuhegger A Steidle P Hutzler C Langebartels L Eberl A Hartmann and F B Dazzo 2004 In situ calling distances and high population independent rhizobacterial cell to cell communication poster International Congress Rhizosphere 2004 Munich Germany M G Forero F Sroubek and G Cristobal 2004 Identification of tuberculosis bacteria based on shape and color Real Time Imaging 10 251 262 M Stelzer and H Reber 2004 Combined methods of image and cluster analysis to estimate the structural diversity of fungal communities Biol Fertility Soils 42 10 16 Howgrave G
54. This advanced CMEIAS shape classifier uses a series of pattern recognition algorithms optimized by us Liu et al 2001 Microbial Ecology 41 173 194 and 2001 Microbial Ecology 42 215 to automatically classify each microbe into one of 11 major morphotypes distinguished by a hierarchy of shape characteristics outlined in Fig 28 This set of morphotypes equals the richness in morphological diversity represented by 98 of the genera in the g h Ed Bergey s Manual of Determinative Bacteriology 55 a Morphotypes classified by CMEIAS Cocci C Spirals S Curved Rods CR U Shaped Rods UR Regular Rods RR Unbranched Filaments UF Clubs CL Ellipsoids E Rudimentary Branched Rods RBR Prosthecates P Branched Filaments BF To produce a CMEIAS 2 morphotype classification CMEIAS must first extract the required shape attributes 7 selections enclosed within a framed area of the Measurement Feature tab page see Fig 6 from each object in an object analysis of the image and then applies these shape analysis data to the pattern recognition algorithms to perform the supervised classification of each object s morphotype 56 7 2 1 amp Fig 28 Hierarchy of characteristics for microbial morphotypes classified by CMEIAS The class labels and pseudocolor assignments against a black background for each morphotype are also indicated The pseudocolored objects may be brighter on screen than in print Morphotype Classif
55. X phase contrast objective lens e g 112 is 0 17 mm thick Check for this information on the objective itself 6 Vary the sample volume as needed to completely fill the volume beneath the coverslip with the edges remaining dry Any excess fluid volume on the slide outside the coverslip should be wicked into a forceps held small piece of filter paper The sample fluid volume under the coverslip will be absorbed within a few minutes by the rehydrating agarose gel layer with no free fluid remaining Prepare only one slide at a time and store it horizontally in a portable humidity chamber until examined microscopically 3 4 Phase Contrast Microscopy of Refractile Immobilized Cells 1 Because image analysis requires a high quality primary image strict adherence to the principles of Kohler illumination with proper phase condenser alignment and uniform background illumination is essential Consult your microscope user manual for detailed instructions to achieve this 2 Optional recommend Introduce a narrow band pass green 546 nm interference contrast filter e g Omega Optical XF1020 beneath the phase contrast condenser to increase resolution by reducing chromatic aberration The increased contrast and improved quality of the grayscale image resulting from use of this filter is well worth its cost Also digital images with transmitted illumination acquired by a CCD camera may require removal of a central bright spot by introducing an infrar
56. XVII Coccus Other 1 Spiral Other Curved Rod Other 3 U shaped Rod C Other 4 Regular Rod Other 5 Unbranched Filament Invalid Object Ellipsoid C Club gt Prosthecate Rome C Rudimentary Branched Aod Branched Filament Cancel plug in C Program Files Uthscsa ImageTool Plug Ins objlabel dll Analysis Object Analysis Edit Classification Results Pause Table 6 in the tutorial worksheet indicates the object classification data containing the cell counts for each morphotype found in both community images Note that we have included 2 columns of classification data for community B before and after editing and have highlighted how the classification data have been changed The frequency counts in the edited results indicate that the regular rod class has increased by 1 and the coccus class has correspondingly decreased by 1 Pause Now view in the tutorial worksheet how these classification data can be used to compare the morphological diversity of the two communities Table 7 is a descending sort of the Table 6 results These data are used to produce Figs 10A and 10B which illustrate two common ways to plot and compare the morphological diversity data acquired by CMEIAS Fig 10A is a vertical clustered Ranked Abundance plot and Fig 10B is a vertical sequential stacked bar graph that indicates the total and relative percentage abundance of each morphotype in the 2 community images respe
57. Zoom out shortcut 35 5 3 Image Transformations The image can be rotated to fit on screen within the ImageTool workspace by selecting Processing gt Transformations gt Rotate 90 180 270 reverse or flip The annotated number and Centroid X Y coordinates for each object will differ in these transformed image orientations see 4 2 3 4 Show Object Numbers on Original Image 5 4 Adjust Image Contrast Brightness Use this ImageTool option to interactively manipulate these features directly on an active 8 bit grayscale image using slider bars before the image is thresholded When needed activate this Brightness Contrast tool by selecting Window gt Show Contrast Control clicking the F7 hotkey or clicking the Brightness Contrast Tool shortcut icon Fig 15 Contrast B rightness ZT T Inverted Negative Image Linear Stretch Fig 15 Adjust Image Contrast Brightness This ImageTool feature is used to adjust contrast and brightness in 8 bit grayscale images plus convert them to the negative images 36 5 5 Negative Image Transformation Checking the Inverted Negative Image box in the Contrast Brightness window Fig 15 will invert the grayscale brightness table for the active image converting it to the corresponding negative image This feature inverts the brightness value of each pixel in the image to its inverse value in the 256 step grayscale values calibration Fig 16 illustrates the
58. able at lt http www astrofegia com Music Wav gt It is important to note that this macro assumes full control of ImageTool while it s running allowing you only to proceed forward accomplished by clicking the OK pushbutton or your keyboard Escape key No option exists to go back to earlier steps in the tutorial minimize ImageTool to access the desktop or access any ImageTool menu items shortcut commands that are not coded steps of the macro itself However you can access all functions on the taskbar Start menu maximize or launch other software applications etc while running this tutorial This is why you should start the other associated applications before starting this training tutorial Click the Escape key repeatedly if you want to skip steps without performing selected tasks and or quickly Il reach the end to close this tutorial macro For general information the paths directed by the command actions are included where first indicated in this manual The Command actions in the macro code activate tasks normally selected from the main menu or the shortcut toolbar These actions run automatically in the training tutorial and do not require user activity Here we have added the text that indicates the macro path activated by the command action where first used e g Command Preferences activates Settings gt Preferences We hope you benefit from this CMEIAS training tutorial especially if you intend to use CMEIAS in you
59. acked 2 d column plot showing the proportional abundance in each size area class from CMEIAS analysis of representative images of communities A and B Cell Size Area Distribution Cell Size Class Area um E lt 0 38 E gt 0 38 0 50 o gt 0 50 0 81 o gt 0 81 0 97 E gt 0 97 1 32 E gt 1 32 3 33 E gt 3 33 5 63 o gt 5 63 6 16 E gt 6 16 6 49 E gt 6 49 7 23 o gt 7 23 7 46 E gt 7 46 8 18 E gt 8 18 8 97 E gt 8 97 10 50 Community A Community B 67 Tables 2 and 3 Ecological statistics of CMEIAS Morphotype Classification Data Shown are the morphological diversity and similarity coefficient indices for Communities A and B derived from CMEIAS morphotype classification data analyzed in EcoStat MorphotypeRichness s5 n Simpson Dominance 0 641 02m Simpson Diversity 0359 0759 MaxofD ows om Inverse dominance 1560 4152 Maxofd s121 11692 devenness oas oss Hmax nes 23 Brllouinmax 1851 231 msni f 2r sone mrsm aseo as Jaccard coefficient 0455 Hormindex ozs 0663 Sneath Sokal distance BrayiCurtis distance 0 512 Chord distance 0 769 CMEIAS EcoStat analyses indicate that communities A and B have 48 82 proportional similarity in morphological diversity with Community B being 2 2 fold higher in morphotype richness and diversity indices and 1 5 fold higher in distribution of morphot
60. age Origin of Coordinates Initial Zoom Setting select the appropriate measurement attribute s Settings gt Preferences gt Measurement Features specify the decimal precision to report data load and spatially calibrate the image and find the foreground objects in the image using the interactive brightness threshold procedure See 4 1 Measurement Features Used in Object Analysis 4 2 Find Objects Settings and 4 3 Image Settings for descriptions of these routines and recommended settings 6 2 Object Counting ImageTool provides 6 different ways to obtain the number of foreground objects in the image Five of these require thresholding and are counted automatically and one is counted manually without requiring a thresholding step 6 2 1 Automatic Object Counting Fig 22 illustrates 3 of the 5 ways ImageTool displays the object count automatically First if Settings gt Preferences gt Find Objects gt Show object numbers on original image is selected 4 2 3 4 Show Object Numbers on Original Image the object count will equal the largest annotated number assigned to the highest positioned object found in the image object 70 in this example see arrow This method is useful if neighboring objects do not obscure the largest numbered annotation 45 Second if Settings gt Preferences gt Find Objects gt Show object count in a message box is selected 4 2 3 1 then immediately after thresholding ImageTool will
61. ageTOool 1 28 ccceceeeeeeeeeeeeees XXVI APPENDIX VI Studies using CMEIAS ceeeseeeeeeeceeeeee eee eeeeeeaeseeeesaeeeeaeee eens XXVII CMEIAS Ver 1 28 Center for Microbial Ecology Image Analysis System Michigan State University Software License Agreement By downloading and installing a copy of the CMEIAS Software and Documentation you agree to the following terms Notification of Copyright CMEIAS is a proprietary product of Michigan State University MSU and is protected by copyright laws and international treaty You as End User must treat CMEIAS like any other copyrighted materials Copyright laws prohibit making copies of the Software for any reason You may make copies of the Documentation for use with a licensed version of the Software however MSU notifications of copyright must be left intact If you have any questions concerning this agreement please contact the Office of Intellectual Property MSU East Lansing Michigan 48824 U S A 517 355 2186 UTHSCSA ImageTool Code CMEIAS incorporates source code from ImageTool software developed by the University of Texas Health Science Center at San Antonio The ImageTool code contained in CMEIAS is a modified version of the code and not the original UTHSCSA ImageTool distributed by the University of Texas The license pertaining to the UTHSCSA code is included in the download installation of ImageTool lt http ddsdx uthscsa edu di
62. al Research in Row Crop Agriculture Michigan State Univ East Lansing MI Matsuyama J M Fukuda S Nakano and F B Dazzo 2002 Ecological roles of protists in microbial loop in streams 67th Annual Mtg Japanese Society of Limnology Fuchu Tokyo Dazzo F B J Liu A Prabhu C Reddy M Wadekar R Peretz R Bollempalli D Trione E Marshall J Zurdo H Hammoud J Wang M Li D McGarrell A Gore J Maya Flores S Gantner and N Hollingsworth 2003 CMEIAS v 3 0 Integrative software package to strengthen microscopy based approaches for understanding microbial ecology 2003 Annual Mtg Long Term Ecological Research in Row Crop Agriculture Michigan State Univ East Lansing MI Dazzo F B A R Joseph A B Gomaa Y G Yanni and G P Robertson 2003 Quantitative indices for the autecological biogeography of a Rhizobium endophyte of rice at macro and micro spatial scales Symbiosis 34 147 158 Reddy C K Feng I Liu and Frank B Dazzo 2003 Semi automated segmentation of microbes in color images In Color Imaging VIII Processing Hardcopy and Applications Proc International Society for Electronic Imaging SPIE 2003 R Eschbach amp G Marcu eds 5008 548 559 Matsuyama J M Fukuda S Nakano and F B Dazzo 2003 Abundance composition and bacterivory of protists on the pebbles in a stream environment 8 th Intn l Congress of Ecology Seoul Korea Dopp H F Dazzo E J Park and A J Smucker
63. alibrate the image before analysis 1 Open the image and zoom in to display an object of known length e g the 10 um magnification bar scale embedded in a remote corner of the image Fig 17 at the highest magnification that it can be fully displayed on screen 2 Select Settings gt Calibrate Spatial Measurement from the main menu This will display a dialog box Fig 17 instructing you to draw a line of known length When placed on the active image the cursor will change to a pencil 3 Define the line by carefully positioning the pencil point at one end of the bar then click and drag the mouse cursor across the length of the bar and release the cursor when the pencil point is located precisely at its opposite end Alternatively left click once with the pencil point at one end of the bar scale and double click with the pencil point at the other end First time users should practice doing this a few times to become skilled and accurate You cannot calibrate with a multi segment line 4 The Calibrate dialog box will automatically display the number of pixels for the line drawn Select the units of measurement e g microns um or 10 meter renamed as micrometer in ImageTool v 3 0 and then enter the bar scale length in this example 10 0 in the input field Fig 17 The units must be chosen before the length value can be entered 5 Click the OK button after you have confirmed the input or click Cancel and repeat if n
64. assification This requirement poses no problem for skilled microscopists using research quality optics image acquisition and processing software and a modern computer Several image archiving softwares are available to help manage large image data sets 3 6 Image Size and Pixel Resolution When an image acquired using a high resolution digital camera is opened in Adobe Photoshop ver 5 or higher the user can check its dimensions by selecting Image Size under the Image menu For example Fig 4A is a screen shot of the Image Size entries for an image acquired using Diagnostic Instrument s Spot software and camera and opened in Adobe Photoshop Ce Original Image NNA Pixel Dimensions 13M 4 Width EE pixels Height i023 pixels Document Size Width fia 264 2c inches Height 14 347 inches Resolution re pixelszinch I constrain Proportions M Resample Image Bicubic Cancel Auto e Peer Resized Image BEL Pixel Dimensions 13M Width 1315 pixels Height 1033 pixels Document Size Width pl ee inches Height 2 928 inches Resolution 263 pixels inch M Constrain Proportions Resample Image Bicubic Cancel Auto Fig 4AB Image Size Image gt Image Size preference box in Adobe Photoshop displaying the default settings of an image acquired by a digital camera e g Spot 2 Diagnostic Instruments See text for instructions how to reduce the image s
65. assification file previously recorded for the same measurement attribute and saved in the Uthscsa ImageTool Calibration folder and click Open ImageTool prompts you to save any new set of upper class limit values as an option This feature automatically fills the appropriate input fields of the Maximum Value in Class with the upper class limits values Note the use of ocd files in CMEIAS v 1 28 8 Verify that the Attribute selection hasn t been changed occurs when loading a calibration file made for another attribute and then click OK to run the object classification The pseudocolor classification image and classification data will display onscreen Fig 27 For example to classify the 3 common morphotypes of cocci straight regular rods and unbranched filaments select the width length shape attribute and enter the 2 optimized upper limits 0 0625 and 0 5 in the first two input fields of the Maximum Value in Class The 3 bin n 1 will include all objects whose attribute value is higher than the n here width length ratio of gt 0 5 upper class limit value entered This example is used in the training tutorial included in the CMEIAS 1 28 download 7 2 CMEIAS 2 Morphotype Classifier A second CMEIAS hierarchical tree classifier uses an optimized subset of 14 shape measurement features to analyze complex microbial communities containing greater morphological diversity than ever before possible see example community below
66. at the bottom of the listed data extracted from the previously analyzed image Caution when concatenating object analysis data you must keep track of the cumulative number of objects from each of the consecutive images within the dataset This is because as indicated in Fig 10 the Results window of IT ver 1 27 only reports on the first 498 objects found 31 T Place object count in results window i Deselect to concatenate Results Std Dev Fig 11 Example of concatenated object analysis data listed in the ImageTool Results window The data extracted from foreground objects in the most recently analyzed image are listed just below those in the previously analyzed image in this case the previous image contained 43 objects The cumulative number of objects analyzed in the data set is indicated in the cell of the leftmost gray column that also contains the Mean Standard Deviation row labels and the last object row displaying its measured values e g 45 in Fig 11 When the total concatenated number of objects approaches the 498 upper limit remember to cut Edit gt Cut Results not copy these data from the ImageTool Results window and paste them into a compatible spreadsheet program so you don t lose data by exceeding this 498 upper limit of objects analyzed Then continue performing object analysis on additional images in the same data set and repeat this data transfer procedure as necessary Note mean std
67. atio Based on that information two horizontal lines have been introduced to indicate the optimized upper class borders of 0 5 and 0 0625 W L that will efficiently separate the three different microbial morphotypes present in this community image Pause Our next step is to perform an IT CMEIAS 1D object classification of this image using these 2 optimized upper class borders of the cell W L attribute These bin limits will satisfy the requirements needed to classify the morphotype of each microbe in this image as being a coccus regular rod or unbranched filament Pause In the Settings Preferences Object Classification tab page select the Report on classifications using a single measurement feature Value Range Number of objects in class and Display new image showing objects colored by classification Deselect the Mean and Standard Deviation values in all classes Then click OK to return to the Object Classification tab page make these selections followed by Apply and OK Object Classification Attributes to send to results window f Report on classifications using a single measurement feature IY Value range classification Mean value of all objects in class IY Number of objects in class Std dev of all objects in class Report on EMEIS Morphotype Classifier using multiple measurement features T Report on objects IY Display new image showing objects colored by classification command preferen
68. bjects routine in image analysis For manual object counting purposes cells touching 2 edges of the image e g bottom and right should be counted whereas those touching the other 2 edges should be excluded from the count 4 2 2 4 Exclude background If checked ImageTool will attempt to identify the region of the image that is background and will not consider it an object in automatic object selection ImageTool considers an object to be background if its width and height are at least 90 of the image size Use this feature setting when analyzing images using CMEIAS plugins 4 2 2 5 Maximum of objects This feature in ImageTool v 1 27 sets the upper limit for the number of objects found by thresholding the active image Unfortunately a coding error prevents the ImageTool s Results window from displaying more than 498 rows So if 499 or more objects are found in an image the Results window will only report on data extracted from the first 497 objects plus the last object found Data on the latter object will be located in the row designated for object 498 This display problem is illustrated in Fig 10 where a low magnification image containing 1 148 objects of bacteria was analyzed 29 Fig 10 Display bug of object analysis data in the Results window This ImageTool v 1 27 bug occurs when more than 498 objects are found in the image s area of interest This problem is a bug in the Image Tool v 1 27 code rather than by
69. cal and genome wide responses of Burkholderia xenovorans LB400 to PCB mediated stress Appl Environ Microbiol 72 6607 6614 Kakizaki T N Hamada S Wada T Funayama T Sakashita T Hohdatsu T Sano M Natsuhori Y Kobayashi and N Ito 2006 Distinct modes of cell death by ionizing radiation observed in two lines of feline T lymphocytes J Radiation Res Online ISSN 1349 9157 Dethlefsen L and T Schmidt 2007 The performance of the translational apparatus varies with the ecological strategies of bacteria J Bacteriol 189 3237 3245 Dazzo F B D Nasr I Leader C Monosmith C Gross R Verhelst T Marsh and C Holzman 2007 Perturbations in microbial community structure associated with bacterial vaginosis measured by CMEIAS computer assisted microscopy and digital image analysis Poster 1 9 2nd Research Forum MSU Center for Microbial Pathogenesis March 29 2007 Eichorst S A J Breznak and T Schmidt 2007 Isolation and characterization of soil bacteria that define Terriglobus gen nov in the phylum Acidobacteria Appl Environ Microbiol 73 2708 2717 Eichorst S A 2007 Isolation and characterization of members of the Phylum Acidobacteria from soils Ph D Dissertation Dept of Microbiology and Molecular Genetics Michigan State University East Lansing Michigan Dazzo F B J Liu G Tang G Zhu C Gross C Reddy C Monosmith J Wang M Li A Prabhu D Nasr C Passmore L Shan
70. ces Pause Next input the parameters for the IT CMEIAS ID classification of this image Click the Attribute drop down list box and select the Width Length feature Then click the Load button navigate to display the contents of the ImageTool Calibration folder and double click to open the OptimizedWidthLength ocd file This calibration file will automatically enter the upper border limits 0 0625 for the Ist 0 5 for the 2nd in the first two input fields This scheme will classify the microbes into 3 bins the Ist group has a W L ratio of up to 0 0625 unbranched filaments the 2nd has a W L ratio between gt 0 0625 and 0 5 regular rods and the 3rd has a W L ratio greater than 0 5 cocci After entering these parameters click the OK button You will then be asked if you wish to save the classification file and current data in the Results window from the previous analysis Answer No for both of these The IT CMEIAS 1D object classifier will then run and report the new data in the Results window plug in C Program Files Uthscsa ImageTool Plug Ins objclass dll Analysis Object Analysis Object Classification Pause Your new classification image should look like Fig 4 in the worksheet Notice that each microbial cell has been pseudocolor coded according to the IT CMEIAS 1D width length class to which it belongs red unbranched filaments class 1 green regular rods class 2 blue cocci class 3 The object counts in each clas
71. clicks the Done command button Fig 30B this interactive editing routine produces an automatic update of the corresponding morphotype frequencies in the contingency table of classification data within the Results window A significant effort was made in development of this CMEIAS edit plug in module to select the pseudocolors that can be rapidly distinguished from one another in the same image and are easily associated with the corresponding morphotype so that interactive recognition of the morphotype classification results becomes almost instantaneous with user experience Fig 29D E The steps to use this CMEIAS Morphotype Classification Edit module for the main Type 1 classification error are illustrated in Fig 30 In this example the CMEIAS 2 classifier automatically assigned a coccus morphotype red pseudocolor to the slightly elongated coccobacillus microbe in the center of the image white arrow because of its high values of roundness and circularity plus length to width ratio of less than 2 1 Fig 30A Such microbial morphotypes are common in unamended soil and oligotrophic aquatic habitats Also Azotobacter chrococcum has this morphology The classification edit feature was then used to reassign its morphotype class to a regular rod blue pseudocolor Fig 30B 30C and when done the morphotype classification frequency count data were updated with the corresponding changes incorporated Fig 30D cocci 62 61 regular rod
72. colored pixels surrounding every object found in the thresholding routine see 5 7 Find Objects by Brightness Threshold Segmentation It does so by walking around the edge of the object finding pixels that have at least one neighbor of a different color or grayscale brightness As with the object numbering see 4 2 3 4 Show Object Numbers on Original Image these annotations are present in a transparent overlay and not made to the image itself Use the print screen key and Windows clipboard to include them in a screen capture image of the monitor display as described above This feature is helpful when optimizing the Minimum Size of Objects pixels see 4 2 2 6 feature to identify which objects in the thresholded image have been found and will be included in the analysis and classification and or to determine if further image editing is needed to fill object holes or accurately find the object s contour while excluding invalid objects and or background noise from analysis 4 2 3 7 Choose Color Double click this rectangular command button to select the color used for the annotated object outlines see 4 2 3 6 Show Object Outlines on Original Image A color palette will display providing the color choices available for selection The color you select can be the same or different than the one assigned to annotate the object number We suggest the bright magenta color illustrated in this above example of a regular rod bacterium 33 4 3 Ima
73. ctively clearly illustrating the similarities and differences in morphological diversity between these 2 microbial communities Pause These same classification data can be used to compute various indices that further compare and contrast the morphological diversity between microbial communities Tables 8 and 9 present the computed results for communities A and B and their interpretation Pause Gain a sense of how CMEIAS image analysis provides several new opportunities to strengthen microscopy based approaches for understanding microbial ecology This ends the CMEIAS 1 28 training tutorial Enjoy CMEIAS XVIII APPENDIX Il CMEIAS Object Analysis Macro CmeiasObjectAnalysis itm If this is the first time you are using CMEIAS we recommend that you perform the exercises in the CMEIAS 1 28 Training Tutorial Macro Appendix I beforehand using the images provided in the program download and installation You must have administrator rights to run this macro file If you want to view the data while they are being extracted from objects in your image optional recommended then before starting this macro you should maximize the Results window position its worksheet near the right edge of the ImageTool workspace and adjust its size to display 3 4 columns and expanded to the full viewable height of your graphical user interface Pause You can use this macro to help guide you through the steps used to perform an object analysis on your
74. cts and perform the object analysis of cell shapes Spatial calibration of the image is not necessary here since all shape attributes are dimensionless measurements Click OK to do these steps open C program files Uthscsa ImageTool Help Community a tif command find objects Pause C Program Files Uthscsa ImageTool Plug Ins objanal dll Pause Maximize the tutorial worksheet at this point to view the results of this object shape analysis for the first 39 objects in Table 5 Consult the CMEIAS 1 28 User Manual or Cmeias128help chm help file to see how the shape measurement features labeled in each column are computed XV Pause ImageTool will always ask if you want to save the current object analysis data in the Results window before they are overwritten with object classification data and visa versa The object analysis data of shape attributes required to perform a CMEIAS morphotype classification are usually not saved unless additional measurement features e g cell area length etc are also included This is because in this example these shape measurement data are only used to compute the pattern recognition algorithms of the CMEIAS morphotype classifier Pause Now click OK to perform the CMEIAS 2 morphotype classification of the microbes in the active image plug in C Program Files Uthscsa ImageTool Plug Ins objclass dll Pause Voila Notice in the new classification image copied to Fig 8 in the tutorial worksheet
75. d S Gantner 2009 Rhizosphere Jn M Schaechter ed Encyclopedia of Microbiology pp 335 349 Oxford Elsevier Dazzo F B J Liu A Jain G Tang C Gross C Reddy C Monosmith A Prabhu R Peretz G Zhu J Wang M Li N Philips A Baruti R Longueuil C Meyers D Nasr I Leader S Zamani C Passmore L Doherty S Dixon P Smith D McGarrell S Pierce S Gantner S Nakano A Smucker E Polone A Tondello A Squartini Y Yanni and R Hollingsworth 2009 CMEIAS v3 1 Advanced computational tools of image analysis software designed to strengthen microscopy based approaches for understanding microbial ecology at single cell resolution 2009 All Investigator Meeting MSU Long Term Ecological Research program May 5 2009 Kellogg Biological Station Hickory Corners MI Gross C A C K Reddy and Frank B Dazzo 2009 CMEIAS Color Segmentation an improved computing technology to process color images for quantitative microbial ecology studies at single cell resolution Microbial Ecology DOI10 1007 s00248 009 9616 7 printed journal version 2010 Microbial Ecology 54 2 400 414 Kui Xian J F Chi MFYang SH Shen YX Jing FB Dazzo amp HP Cheng 2010 Movement of rhizobia inside tobacco amp lifestyle alternation from endophytes to free living rhizobia on leaves J Microbiol Biotech 20 2 238 244 2009 online DOI 10 4014 jmb 0906 06042 XXXIII
76. d Slides ccceceeeeeeeeeeeeeeeeeeeee nese eeeeeeeeeeeeeeees 14 3 3 Preparation of Dispersed Microbial Samples ccccesceeeeeeeeeeeeeeeseeeeeeeeaeees 15 3 4 Phase Contrast Microscopy of Refractile Immobilized Cells 15 3 5 Image Acquisition amp Sampling Density cccceeeeeeeeeeee seen seen eee eee eeeeeeeeeeeeees 16 3 6 Image Size and Pixel Resolution ccccceseeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeseaeeeeeseeeees 17 3 7 mage Editing soja eics ous tuimutadracaasans E REE a dias coectn ena ERE EERE 18 3 8 Adding a Calibrated Bar Scale n nnsaesnesnusnnnnnnrnunnnnnnnnnnnnnnnnnunnnnnnnnunnnnnnnnnnnnnnna 20 4 SETTING PREFERENCES PRIOR to IMAGE ANALYSIS 0 ccceeeeeeeeeeeeeeeeeeeeee 21 4 1 Measurement Features Used in Object AnalySis cccceeeeeeeeeeeeeeeeeeeeeeeeeeees 21 4 1 1 Measurement features preference Page 2 2 2 2cccceceecececececeenececeeeeenensaeesenesesenneecesaes 21 4 1 2 Definitions of measurement features for object ANAlySIS 2 2 10ccccccenceeenececseeeeeeeeennens 22 4 1 3 Measurement Precision decimal plaCeS 111cccssssescnsceseecanecensnaneeeesnecssaaaneeaaanseses 26 4 1 4 Display of object analysis data in the Results WiINKOW c1ceccececeseneuseceeuneceneneeenenees 27 4 2 Find Objects Sewn GS isc cvisiccistensicvndericsavacuddeveatwansssdenns anaie aeaa ade daii aa EAER 28 4 2 1 Area of Interest AOI OptiOnS
77. d the program CMEIAS an acronym for the Michigan State University Center for Microbial Ecology Image Analysis System CMEIAS Ver 1 28 is not a stand alone program but rather consists of several custom plug ins that operate in the host program UTHSCSA ImageTool Ver 1 28 a free downloadable open architecture software prepared for the PC Two of these CMEIAS 1 28 plugins objanal dll and objclass dll are derived work representing modified plugin versions of object analysis and object classification plugins and not the original UTHSCSA ImageTool distributed by the University of Texas The accuracy and performance of microbial morphotype classification were thoroughly tested with ground truth data using CMEIAS v 1 27 in UTHSCSA ImageTool running in Windows NT 4 0 Liu et al 2001 The software runs using Windows 32 bit operating systems 2000 XPpro Vista and 7 In summary CMEIAS is an accurate robust flexible semi automatic computing tool that fills a major gap by significantly strengthening microscopy based quantitative approaches for understanding microbial ecology at spatial scales relevant to the microbe s niche and can serve as a useful adjunct in the analysis of microbial community structure in situ without cultivation Many examples of CMEIAS usage are listed and hotlinked in the Publications using CMEIAS webpage at our CMEIAS website New features of the CMEIAS v 1 28 upgrade include many user friendly improvements of displayed
78. design and developers of ImageTool have fixed it in ImageTool release versions 2 X and higher CMEIAS v 1 28 users should avoid this problem by setting this field to 498 objects Fig 9 and include no more objects than this upper limit in an image for object analysis Since individual images intended for CMEIAS morphotype analysis should ideally contain no more than 170 microbes anyway to avoid excessive crowding see 3 1 Image Requirements for Image Analysis it may be necessary to dilute the sample acquire new images in microscope fields containing fewer microbial cells redraw the AOI polygon to specify smaller portions of the images and or use the Manually Select Objects mode to avoid exceeding the 498 upper limit see 3 1 Image Requirements for Image Analysis 3 3 Preparation of Dispersed Microbial Samples 4 2 1 2 Search in AOI 5 7 Find Objects by Brightness Threshold Segmentations and 5 8 Manually Select Objects for more details If the number of objects found in an image exceeds whatever limit is specified in this Maximum of objects text box a new message box will display instructing the user to specify a larger number in this field of the Find Objects tab page 4 2 2 6 Minimum object size pixels This input field defines the minimum object size in pixels required to be automatically included in the threshold routine Objects containing fewer pixels than specified will be excluded by automatic object selection Note When optimized
79. e Area x A f ae i t i A e P Classifications a window VA sn P p L Maximum Value in Class _oK Results OF Xx Yalue Range Count Mean Yalue Std Dev a Mean 10 63 6 65 Std Dev 6 27 0 48 Cancel 0 98 Help 2 Open and spatially calibrate the segmented image to be analyzed see Fig 17 and 5 6 Spatially Calibrate the Image 3 Select Analysis gt Object Analysis gt Find Objects and perform the brightness threshold routine see Figs 18 21 and 5 7 Find Objects by Brightness Threshold Segmentation to find all the foreground cells of interest in the image There are 170 microbes in the CocciRegularRodFilament tif image used to make Figs 26 and 27 4 Select Analysis gt Object Analysis gt Object Classification to open the IT CMEIAS 1D Define Object Classifications window Fig 26 53 5 Click the Attribute pull down list box and select the measurement attribute desired Area in Fig 26 6 Enter the upper class limit for each bin in the Maximum Value in Class input fields Fig 26 Alternatively click the Load button select an appropriate ocd classification file previously recorded for the same measurement attribute and saved in the Uthscsa ImageTool Calibration folder and click Open ImageTool prompts you to save any new set of upper class limit values as an option This feature automatically fills the appropriate input fields of the Maximum Value in Class
80. e I bow J quarter circle to pyramid K budding L rod with two encircling rings M N spiked spheroids O star P gull wing Q R coiled unbranched filament 64 The last valid bacterial shape in this illustration is an unbranched coiled filament that forms enclosed loops objects Q and R of Fig 31 resembles cooked spaghetti This is the only morphotype in Fig 31 with both type 2 and type 3 errors requiring the second and third editing tasks to classify correctly CMEIAS regards object Q as two objects one a spiral represented by the outer contour of the cell and the other a coccus represented by its inner contour outer contour of the closed loop These classification errors are corrected by first reassigning the internal loop as an invalid object eliminating the type 3 error assigned a very dark blue pseudocolor that is barely discernible from the black background next reassigning the cell object as an Other 1 morphotype and finally renaming it as a Coiled Unbranched Filament after exporting the classification data to a compatible spreadsheet program 65 CHAPTER Graphics and Ecological Statistics of CMEIAS Object Analysis amp Classification Data This chapter illustrates various graphic plots and tables of ecological statistics derived from CMEIAS data that would typically be included in a microbial community analysis The two images used in this analysis Fig 29A and Fig 30C represent 2 differen
81. e class labels columns indicate CMEIAS classifier assigned labels and numbers of correctly and incorrectly classified cells in each morphotype class For example CMEIAS classified 1212 of 1217 cocci cells correctly 99 6 accuracy and incorrectly classified 4 cocci cells as regular rods and 1 coccus cell as an ellipsoid morphotype 60 7 2 4 Sources of Morphotype Classification Errors amp the CMEIAS Edit Feature Extensive analyses of microbial communities from the bovine rumen anaerobic sludge bioreactors gut fluid of Reticulitermes flavipes termites human dental and tongue surface biofilms legume root nodules and soil indicate that the error rate for the CMEIAS morphotype classifier is 3 Liu et al 2001 CMEIAS makes 3 types of classification errors in images that are properly segmented with adequate pixel sampling density The type 1 error occurs when the cell s shape is at the border between two morphotypes defined quantitatively in 14 dimensional space and the visual classification is not the one assigned by CMEIAS For example CMEIAS is well justified microbiologically and geometrically to classify coccobacilli as cocci Liu et al 2001 but the user may prefer to classify them as short regular rods despite their length to width ratio being less than 2 1 A less frequent type 2 error results when the microbe has a very rare shape that does not match any of the 11 predefined morphotype classes in the CMEIAS 2 morphotype
82. e each of these measurement features Check the boxes for the Length Width and Width Length ratio The latter is a shape measurement feature located within the framed area labeled CMEIAS Morphotype Classifier not to be confused with the Length Width ratio Also for all analyses in this tutorial set the Measurement Precision to 2 decimal places Click OK to return to the Measurement Feature tab page and make these selections followed by Apply and OK Measurement Features CMELAS Morphotype Classifier Other Measurement Features Roundness T Area Aspect Ratio T Elongation Perimeter T Centroid amp 7 Compactness Feret Diameter T Gray Centroid amp T Masinum Curvature IY Length Integrated Density I width Length I width Min Gray Level Area BB Area T Lengthwidth T Mean Gray Level Fourier Descriptors Major Asis Length Median Gray Level Major Axis Angle 7 Mode Gray Level ReSene Te THE TSecian e Minor Axis Length Max Gray Level decimal places T Minor Axis Angle 7 Std Dev Gray Level command preferences VI Pause Click OK to open the CoccusRegularRodFilament tif image open C program files Uthscsa ImageTool Help CoccusRegularRodFilament tif File Open Pause Spatial calibration of an image must precede the measurement of object sizes in order to report the output values in user selected units rather than the pixel default This interactive step is best done while zoo
83. ecessary All dimensional analyses performed on this image will be based on this spatial calibration 38 Useful tips You can remove calibration from an image by selecting the Calibrate Spatial Measurement command drawing a line of any length and accepting its length value in pixels Since shape is a dimensionless characteristic independent of size all shape measurement features required to perform the CMEIAS 2 morphotype classifier are dimensionless and therefore they can be computed in pixel units extracted from objects without the requirement to spatially calibrate the image However you should spatially calibrate the image in order to report object analysis data for measurement features with dimension e g length in the unit desired The Settings gt Load Spatial Calibration command loads previously saved spatial calibrations to be used in dimensional analysis To create this ite spatial calibration file click yes when the dialog box offers to save the calibration of an image that serves as the default magnification for subsequent images Once loaded this spatial calibration becomes the default for all measurements on all images both those currently open and those opened later in the same image analysis work session The loaded spatial calibration remains in effect until the end of the current image analysis session or until another spatial calibration is loaded To revert to the uncalibrated behavior you must load t
84. ed absorbing filter in the light path 3 For morphotype classification analysis use a 100X phase 3 PlanApochromat oil immersion objective if available This type of objective lens is flat field corrected to minimize spherical aberration that produces blurred out of focus regions at the periphery of the captured image 4 The numerical aperture N A of the oil immersion system is a function of the N A of the objective lens and the condenser Since the N A of the PlanApochromat oil immersion objective lens is gt 1 a drop of immersion oil should be applied to the condenser lens beneath the slide to fill the gap between the condenser and the underside of the slide before adjusting for K hler illumination in order to realize the objective lens full N A If the gap between the condenser and the slide is not oiled the highest possible N A is 1 regardless of the N A of the objective lens 5 Find fields of view in which the refractile bacteria are optimally separated from one another Due to edge drying cells close to the edge of the coverslip will be immobilized first Ideal fields to photograph are ones that contain very refractile cells that HAVE JUST BEEN immobilized Bacterial cells will eventually diminish in refractility as the agarose platform continues to absorb water swell further and completely surround them 6 Acquire micrographs from as many different locations as are feasible so that a sufficient number of randomly selected ce
85. ed if Roundness gt 0 8 Referring to Fig 7b a 2 dimensional presentation of a straight rod with rounded ends can be represented by a rectangle attached to a half circle at each of its two poles and its length can be approximated as a b On the other hand we use the Major Axis Length feature to define the length of a more rounded object Thus the CMEIAS formula to compute the object length is as follows 2 Perimeter a 2 J Perimeter 4x Area Length 27 otherwise use Major Axis Length ifRoundness lt 0 8 The algorithms to measure cell length and the major axis length return values that are approximately equal in accuracy for elongated microbes with a straight longitudinal axis e g regular rod However the adaptive algorithm in CMEIAS used for cell length is more accurate than the major axis length for microbes with a curved axis e g spirals bent unbranched filaments U shaped rods curved rods CD in Fig 7A illustrates the problem and represents a significant strength of CMEIAS object analysis 24 The width of an object is defined as its average width along the skeleton and is approximately computed for these two types of objects as follows Perimeter X Perimeter 42 Area Width m otherwise use if Roundness lt 0 8 Area Major Axis Length The ratios Width Lengthand Length Width between the measurements Width and Length calculated using the above formulas a
86. egia com Music Wav Pause This macro only allows you to proceed forward No option exists to undo go back to revisit earlier steps or minimize ImageTool while running this macro Repeatedly click the Escape or Enter key or push the OK or Cancel button to skip steps without performing selected tasks and or quickly reach the end to close this tutorial macro Pause It is assumed that you are already familiar with the operation of the host program UTHSCSA ImageTool ver 1 27 ImageTool provides an operator manual Help files and an object analysis tutorial in its program download to help you learn its operation The Results window should already be opened adjusted in size to display 3 4 columns and expanded to the full viewable height of your graphical interface and moved to the far right side of the ImageTool workspace I displayed on your monitor Maintain this open status of the Results window throughout this tutorial Pause This CMEIAS v 1 28 tutorial will describe how to 1 select the required preference settings 2 perform an automatic object analysis of cells in a community image 3 optimize the upper class limits of bins to perform an IT CMEIAS 1D classification of the morphotype diversity in a simple microbial community 4 compare the cell size distribution of different complex communities using the IT CMEIAS 1D classifier 5 measure the morphological diversity of complex microbial communities using the CMEIAS 2
87. egmentation threshold procedure that finds the foreground objects in the active image 4 2 1 Area of Interest AOI Options 4 2 1 1 Search entire image When checked ImageTool will look for candidate objects in the entire image This option should not be checked if the image contains a magnification bar scale otherwise that invalid object will be erroneously regarded as a foreground object of interest and be included in the object analysis 4 2 1 2 Search in AOI Area of interest When checked ImageTool will ask you to select an AOI during the Find Objects routine This is done before the thresholding stage by placing the cursor on the active image and then using the converted pencil cursor to draw a polygon single click creates a corner that encloses all foreground objects of interest see 5 7 Find Objects by Brightness Threshold Segmentation Double click to connect the first and last corners to close the AOI polygon When the Include objects at edge of image feature is deselected described below only those objects located completely within the select area of this polygon will be included and analyzed Also this Search in AOI feature must be selected when the 28 image contains a bar scale so the polygon can be drawn to exclude it from object analysis see 5 7 1 Activating threshold selections 4 2 2 Search options 4 2 2 1 Manually select objects Check this feature to manually select the objects to be analyzed from t
88. eground microbes of interest open C program files Uthscsa ImageTool Help Community b tif command zoom in command zoom in command spatial calibrate command zoom out command zoom out command find objects XI Pause Click OK to perform the object analysis of the microbes in this image Note that the reporting of data will be CONCATENATED in the Results window below the previous data on the same measurement attributes extracted from the microbes in the Community A tif image plug in C Program Files Uthscsa ImageTool Plug Ins objanal dll Pause Open the tutorial worksheet and view the top portion of Table 3 containing the cell area data for both images plus 2 additional column arrays showing the same data after an ascending sort Pause Figure 5 in the worksheet is a line plot of the sorted object areas in communities A and B Although this straightforward object analysis produces accurate data this display of results reveals only marginal differences in their cell size distribution A more powerful data mining approach to analyze this community characteristic would be to optimize the upper class bin limits in regions where the 2 communities differ most in Fig 5 and then apply these values to an IT CMEIAS 1D classification of cell areas Let s do that analysis to emphasize this point Pause For this object classification we do not need to produce a new image with the objects colored according to their classificatio
89. en shortcut icon far left button in the toolbar or 3 click the F2 hotkey This displays an Open window where you specify the appropriate drive folder and name of the image file to be analyzed and click the Open button The image window must be active title bar colored to perform subsequent steps Consult the ImageTool Operator Manual It doc file to work with stacked or tiled image displays 5 2 Adjust Image Size Use the Zoom In F11 hotkey or Zoom Out Ctrl F11 hotkey toolbar shortcut icon to adjust and display the entire image at the highest magnification that fits within the ImageTool workspace The normal zoom range is 8 1 to 1 4 However to display very large images you can override the 1 4 zoom out limit by using your mouse left button to apply multiple rapid repeated clicks of the Zoom out icon For example the very large image in Fig 14 a 416 inch wide montage image built from 26 geo referenced scanning electron micrographs of bacteria colonized on a rice root was zoomed out to 1 41 by this neat little CMEIAS shortcut in order to display it fully on screen As computer speeds continue to be developed faster and faster the ability to use this override trick will probably be lost i RootMontagel 11510245448 ottom 1 41 Feo et a ee Bt as Pat R ote mi e Aa e ee z ihe Fig 14 Adjust image size A very large image 416 inches wide displayed on screen by rapidly repeatedly clicking the toolbar
90. er when used in conjunction with computer assisted image analysis There are two main advantages of using digital image processing and pattern recognition techniques in conjunction with microscopy for quantitative studies of microbial ecology First automatic image analysis reduces the amount of tedious work with microscopes needed to accurately quantify in situ morphological diversity abundance and metabolic activity of microbes Secondly these techniques provide an important quantitative tool that can significantly enhance the polyphasic analysis of the diversity abundance functions and spatial features of complex microbial communities in situ without cultivation One of the most important and yet most tedious tasks performed during microscopical analysis of microbial communities is the classification of observed cells into known morphological categories and recognition of new categories as well if new distinct characteristics are captured Inclusion of morphological diversity in evaluations of microbial community structure is more useful and valid if the cells are actively growing rather than in a non growing quiescent state since the latter is more commonly associated with pleomorphic dwarf cells This is because distinctive cell morphologies reflect the phenotypic expression of complex networks of genes involved in the synthesis and maturation of the shape determining murein sacculus plus other genes dedicated to the cell division cycle that are
91. expressed only during active growth A major challenge in microbial ecology is to develop reliable and facile methods of computer assisted microscopy that can analyze digital images of complex microbial communities at single cell resolution and compute useful quantitative characteristics of their organization and structure without cultivation Although several image analysis systems can classify microbes according to their cell sizes automatic classification of cells according to their distinctive morphology a dimensionless characteristic based on several shape features represents a much more challenging task Most commercial image analysis systems include some shape measurement features that compute the roundness or circularity of cells and these characteristics are sufficient to distinguish regular rods and cocci the most common morphotypes of bacteria However the difficulty increases with morphological diversity since automatic classification of most other microbial morphotypes requires measurement of multiple shape and size features to resolve the distribution of their morphological space Some custom image analysis systems are adequate for automatic morphotype classification of spheres straight rods and vibroids or prolate spheroids This represents the morphological diversity of some marine bacterioplankton communities However comprehensive image analysis systems capable of automatically classifying much broader morphological diversity in
92. f objects found displayed when specified in the Find Objects tab page see 4 2 3 1 Show Object Count in a Message Box Al 5 7 3 Find objects in a non binary image Eix B i Kbsit24 c 1 1 A Y a e D N 7 Fig 20A D The various steps to find the foreground objects of interest in a non binary grayscale image using the ImageTool brightness threshold segmentation procedure Nice image lt A anat 7 If the grayscale image contains pixels of varying brightness Fig 20A select the Manual thresholding method in the Find Objects dialog box Fig 18 and click OK Draw the polygon enclosing all foreground objects of interest see section 5 7 2 Find objects in a binary image while excluding invalid objects and the bar scale if present Closure of the AOI polygon will automatically fill it in red and open an ImageTool window displaying a frequency histogram of the gray levels within the image Fig 20B When positioned over the slider bar the cursor changes from a white arrow to a black plunger For positive images dark foreground objects against a bright background see 5 5 Negative Image Transformation slowly slide the right most plunger to the left causing pixels with brightness gray level values that fall between the two endpoints to become red while pixels outside the range background will revert to their original gray levels Continue sliding the plunger position furthe
93. g itdesc html gt and this operator manual and is binding on you upon your use of the CMEIAS code License MSU grants End User the royalty free non exclusive non transferable right to use CMEIAS Ver 1 28 software for research training or educational purposes You may not redistribute transfer rent lease sell lend sub license prepare derivative works decompile or reverse engineer the CMEIAS Software without prior express written consent of MSU at the above address MSU retains title to CMEIAS including without limitation the Software and Documentation End User agrees to use reasonable efforts to protect the Software and Documentation from unauthorized use reproduction distribution or publication All rights not specifically granted in this Agreement are reserved by MSU Warranty CMEIAS Software and Documentation are provided as is MSU MAKES NO WARRANTY EXPRESS OR IMPLIED TO END USER OR TO ANY OTHER PERSON OR ENTITY SPECIFICALLY MSU MAKES NO WARRANTY OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE OF CMEIAS SOFTWARE OR DOCUMENTATION MSU WILL NOT BE LIABLE FOR SPECIAL INCIDENTAL CONSEQUENTIAL INDIRECT OR OTHER SIMILAR DAMAGES EVEN IF MSU OR ITS EMPLOYEES HAVE BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES IN NO EVENT WILL MSU LIABILITY FOR ANY DAMAGES TO END USER OR ANY PERSON EVER EXCEED THE FEE PAID FOR THE LICENSE TO USE THE SOFTWARE REGARDLESS OF THE FORM OF THE CLAIM General If any provisi
94. g microbial communities e Automatic morphotype classification of complex communities exhibiting high morphological diversity will require development of a more flexible and robust computer assisted image analysis system than those currently available e Phase contrast light microscopy of dispersed samples immobilized on agarose coated slides is a simple yet effective direct method to acquire images with the resolution and range of object brightness at high magnification that are sufficient to reveal the rich morphological diversity of actively growing microbial communities Its essential requirements to detect microbes are that their size exceeds the 0 2 um limit for light microscopy their refractive index differs from that of the surrounding medium and high quality optics are available to acquire the images with sufficient resolution to accurately define the foreground cells contour against the background We have improved on existing image analysis systems of computer assisted microscopy by introducing new measurement features and robust object classifiers capable of automatically classifying most of the predominant microbial morphotypes encountered in digital micrographs of complex microbial communities growing in nutrient enriched habitats and have implemented these features in a flexible user friendly and robust semi automatic image analysis system designed to strengthen microscopy based methods for understanding microbial ecology We name
95. ge Settings 4 3 1 Origin of X Y Cartesian coordinate system in image windows Image Origin of x Y Cartesian coordinate system in image windows Upper left x Y origin is 0 0 Lower left x Y origin is 1 1 Prompt to save untitled images T Show full pathnames in image title bars Initial Zo0om Scale Facto C 8 1 C 41 C 21 C 11 1 2 C 13 C 1 4 Fig 13 ImageTool s Image tab page Settings gt Preferences gt Image contains the settings that control the manner in which images are displayed The locations of pixels in an image are defined by their unique Cartesian x y coordinates relative to a corner landmark origin ImageTool s default origin is the Upper left corner positive Y values go down The coordinate of the origin point is controlled by the X Y origin is 0 0 and X Y origin is 1 1 option buttons These selections control how the x y coordinates of the mouse cursor on the image display in the status bar at the bottom of the ImageTool workspace and for reporting object analysis results using the object centroid and gray centroid measurement features When using CMEIAS to extract georeferenced data from objects set the landmark origin of coordinates to the Lower left corner positive Y values go up and the origin point to X Y origin is 0 0 as illustrated in Fig 13 4 3 2 Prompt to save untitled images This ImageTool feature prompts you to save each new image created by the image proces
96. goleva B Yu and A Jain 2001 CMEIAS a computer aided system for the image analysis of bacterial morphotypes in microbial communities Microbial Ecology 41 173 194 and 42 215 Tiedje J A Fernandez S Hashsham S Dollhopf F Dazzo R Hickey and C Criddle 2001 Stability persistence and resilience in anaerobic reactors a community unveiled Advances in Water and Wastewater Treatment Technology Matsuo Hanaki Takizawa Setoh eds pp 13 20 Elsevier Amsterdam Dollhopf S S A Hashsham F B Dazzo C Criddle and J M Tiedje 2001 The impact of fermentative organisms on carbon flow in methanogenic systems under constant low substrate conditions Appl Microbiol Biotechnol 56 531 538 Dazzo F B J Liu F I Liu and G P Robertson 2001 In situ analysis of microbial communities on the 2 day old white clover rhizoplane and the mature BT corn phylloplane using CMEIAS innovative software for computer assisted microscopy Annual Mtg Long Term Ecological Research in Row Crop Agriculture Michigan State Univ East Lansing MI Hashsham S A T L Marsh S L Dollhopf A S Fernandez F B Dazzo R F Hickey C S Criddle and J M Tiedje 2001 Relating function and community structure of complex microbial systems using neural networks Advances in Water and Wastewater Treatment Technology Matsuo Hanaki Takizawa and Satoh eds pp 67 77 Elsevier Amsterdam Yanni Y G R Y Rizk F K Abd El Fattah
97. gration of all or part of the source code into a product for sale or license by or on behalf of Licensee to third parties or 2 distribution of the binary code or source code to third parties that need it to utilize a commercial product sold or licensed by or on behalf of Licensee UT MAKES NO REPRESENTATIONS ABOUT THE SUITABILITY OF THIS SOFTWARE FOR ANY PURPOSE IT IS PROVIDED AS IS WITHOUT EXPRESS OR IMPLIED WARRANTY THE UT SHALL NOT BE LIABLE FOR ANY DAMAGES SUFFERED BY THE USERS OF THIS SOFTWARE By using or copying this Software Licensee agrees to abide by the copyright law and all other applicable laws of the U S including but not limited to export control laws and the terms of this license UT shall have the right to terminate this license immediately by written notice upon Licensee s breach of or non compliance with any of its terms Licensee may be held legally responsible for any copyright infringement that is caused or encouraged by Licensee s failure to abide by the terms of this license For more information contact lt dove uthscsa edu gt Background Information 1 J Liu F B Dazzo O Glagoleva B Yu and A Jain 2001 CMEIAS a computer aided system for the image analysis of bacterial morphotypes in microbial communities Microbial Ecology 41 173 194 and 2001 Microbial Ecology 42 215 http ceme msu edu cmeias 2 Gross C A C K Reddy and Frank B Dazzo 2009 CMEIAS Color Segmentation an improved compu
98. gure 7a Note that this vector can extend outside the boundary of curved objects as illustrated Also note that this measurement feature extracts the true length of elongated objects with a straight longitudinal axis but underestimates the object s length when it has curvature or is irregularly shaped Fig 7AB Vectors of a curved A and straight B rod used to define the measurement features of CMEIAS ImageTool v1 28 from Liu et al 2001 Minor Axis Length The length of the longest line that can be drawn through the object perpendicular to the major axis This size measurement is the vector EF in Figure 7a Elongation The ratio of the length of the major axis to the length of its minor axis i e CD EF in Figure 7a The result is a value gt 1 If the elongation is 1 the object is roughly circular or square The ratio increases from 1 as the object becomes more elongated Filamentous objects have large elongation values except when they contain localized width enlargements important for biofilm measurements 23 Compactness Computed as N 4 Area x Major Axis Length This shape feature measures the object s circularity representing the ratio of the Feret diameter defined below to the object s major axis length and ranges between 0 and 1 Objects with a compactness value of 1 are roughly circular Maximum Curvature The curvature at a point on the boundary of an object is defined as the inverse of the angle at that point
99. he uncalibrated No Calibration itc file located in the Uthsesa ImageTool Calibration folder Fig 2A space left blank intentionally 39 5 7 Find Objects by Brightness Threshold Segmentation The foreground objects of interest e g bacteria in the image must be found before they can be analyzed In ImageTool this is accomplished by threshold segmentation based on differences in grayscale brightness between the pixels that define the foreground objects from those of the background ImageTool considers an object to be background if its width and height are at least 90 of the image size Using this threshold procedure one can select the objects automatically all of this section or manually 5 8 Manually Select Objects Successful thresholding of grayscale images to automatically select objects Fig 9 requires that all the foreground object pixels have brightness values outside the range of those that are background Commonly images require some editing to fulfill this fundamental criterion for digital image analysis see 3 7 Image Editing 5 7 1 Activate threshold selections To find the foreground objects in a user defined Area Of Interest polygon within an image select Settings gt Preferences gt Find Objects gt Search in AOI gt Apply Fig 9 Then activate the threshold routine by selecting Analysis gt Object Analysis gt Find Objects or click the Find Objects shortcut icon to display its dialog box Fig 18
100. hose automatically identified by the Find Objects command While mageTool will still find all of the objects for you it will only analyze those objects you choose Use this feature to select only the segmented objects of interest in an image that also contains many invalid objects found or if the number of objects of interest in the thresholded image exceeds the 498 limit that can be reported on in the Results window of ImageTool v 1 27 see also 4 2 2 5 Maximum of Objects and 5 8 Manually Select Objects 4 2 2 2 Automatically select objects If checked ImageTool will analyze all objects found in the image that satisfy the constraints specified in the next 4 sections This feature is commonly selected when using CMEIAS 4 2 2 3 Include objects at edge of image If checked then automatic object selection will retain and analyze objects that touch any edge of the image or AOI polygon This option is off by default since the size and shape of border objects may be incomplete and therefore erroneous Deselect this feature when using CMEIAS to analyze cell size and morphotype classification Also you must carefully consider whether partial objects located at the edge of the image should be counted You can add a little extra noise free margin of background pixels to an image while editing it image gt larger canvas size in Photoshop to facilitate drawing the AOI polygon and avoid excluding important objects at the edge during the Find O
101. ication find misclassified objects and edit the results if necessary 3 Generate Classification Out file CMEIAS automatically generates and saves a small text file of morphotype classification results in the same directory where the analyzed image file is located This text file contains a numbered list of morphotype label classifications for each individual object 59 corresponding to its annotation number in the thresholded image and is named the same as the corresponding image file followed by an out file extension This out file represents an automatic backup of the object classification result and was originally designed by the developers of CMEIAS for use in ground truth testing of the morphotype classifier These automatically created and saved out files can be deleted in Windows Explorer 4 CMEIAS Morphotype Classification accuracy Extensive testing Liu et al 2001 Microbial Ecol 41 173 194 indicates that the CMEIAS 2 morphotype classifier performs with an overall accuracy of 97 0 on properly edited images indicating that accurate classification of the rich morphological diversity in complex microbial communities is now possible Table 1 aje ite ab JA tel ZE Filament sie E P5 LE Table 1 Confusion table indicating accuracy test results of the CMEIAS 2 morphotype classifier using a large dataset of ground truth images from Table 4 of Liu et al 2001 Rows indicate tru
102. ication Criteria Label Morphotype Pseudocolor Cell L Rounded cells spheroid ovoid coccobacilli A COCCI IL Elongated cells A Unbranched 2 cell poles l Repeated waveform B SPIRAL 2 Single regular curvature a Curvature is vibroid or crescent C CURVED ROD b Curvature is U shaped D U SHAPED ROD 3 Linear no repeated curvature of the medial axis a Constant cell width 1 Length Width ratio is 16 1 E REGULAR ROD 2 Length Width ratio is gt 16 1 F UNBRANCHED FILAMENT b Widest at center tapered at both poles G ELLIPSOID c Wider at one pole 1 Gradually tapered at opposite pole H CLUB 2 Thin stalk appendage I PROSTHECATE B Branched 2 cell poles 1 Length Width ratio 16 1 J RUDIMENTARY BRANCHED ROD 2 Length Width ratio gt 16 1 K BRANCHED FILAMENT 57 7 2 2 Steps to Perform a CMEIAS Morphotype Classification 1 Set Settings gt Preference tab pages as follows Measurement Features tab Fig 6 select all 7 shape attribute check boxes within the Morphotype Classifier frame Roundness Elongation Compactness Max Curvature Width Length Area BB Area Fourier Descriptors Find Objects tab Fig 9 select Search in AOI Automatically Select Objects Exclude background Show object numbers on original image Show object outlines on original image Deselect Include objects at edge of image Object Classification tab Fig 25 select Report on
103. if only shape measurement attributes are included in the object analysis However if you also selected to include size attributes e g length area during object analysis you should spatially calibrate the image with the appropriate unit e g um before the object analysis step Click OK to spatially calibrate the image or push the Escape button twice to skip this step and proceed to the next one Command spatial calibrate Pause Next select the type of brightness thresholding procedure from the dialog box If you previously selected the Search in AOI setting in the Find Objects preferences then follow the instructions to draw a polygon around the region that you wish ImageTool to search Then manually threshold the image to find your foreground objects of interest or let the program find them automatically if your image is binary and you selected None image is already thresholded command find objects Pause Now click OK to activate the object analysis routine that will extract the 14 different shape attributes required to classify the microbial morphotype of each object in the image plus any other selected attributes and report the data in the Results window XXIV plug in C Program Files UTHSCSA ImageTool Plug Ins objanal dll Pause Before running the Morphotype Classifier ImageTool will ask if you want to save these object analysis data in the Results window Answer No if you are only interested in the morphotype
104. imentary Branched Rod Branched Filament 16 pseudocolors distinguish morphotype assignments following object classification IP o Other5 0 Fig 29D 29E The 16 pseudocolors used in CMEIAS morphotype classification displayed on the objects in the classification result image and the classification data in the Results window Some pseudocolors may be more easily distinguished when displayed on the monitor than when printed 7 2 3 Important points regarding the CMEIAS Morphotype Classifier 1 Speed of operation Using a PC with Pentium II operating at 300 MHz with 384 MB RAM and 32 MB video card the algorithms for object analysis and object classification presented their results for most of the images used in CMEIAS development in an average of 2 0 sec and 0 2 sec per image respectively These durations are well below the approximate 10 sec limit of user irritation Most of this time is actually used to produce the graphical display rather than computation per se and of course will be even shorter duration for faster computers 2 Use of classification pseudocolors Rather than just being cosmetic each pseudocolor used in the CMEIAS morphotype classification was selected with great care so it could easily be associated with the specific morphotype class to which each object belongs Once learned this pseudocolor recognition scheme becomes an efficient way to visually inspect the accuracy of the automatic morphotype classif
105. inates of all pixels belonging to the object This attribute reports the unique Cartesian x y coordinates in the spatial unit used to calibrate the image for every object found in the 2 D image relative to the landmark origin selected in the Settings gt Preferences gt Image tab page see 4 3 1 Origin of coordinate system in image windows Note the centroid pixel may locate on or outside the contour boundary of curved or irregularly shaped cells Aspect Ratio This shape measurement attribute is the ratio between the minimum and the maximum distance from the points on the object s boundary to its centroid OB OA in Fig 7a The aspect ratio algorithm will report a value of 0 for curved objects whose centroid lies at their periphery Major Axis Angle Minor Axis Angle Gray Centroid x y Integrated Density Min Mean Median Mode Max and Std Dev Gray Level are measurement features provided by ImageTool that are not used by CMEIAS to classify bacterial morphotype at this time See the ImageTool Operator Manual It doc for information on and use of these other measurement features 4 1 3 Measurement Precision decimal places Measurement Precision decimal places E Enter this number in the Settings gt Preferences gt Measurement Feature tab page Fig 6 to set the decimal places to report the data output of individual objects in the object analysis Results window Since images of bacteria are typically calibrated in micrometers
106. into a Windows compatible worksheet Edit gt Copy Results of all data in the grid to the system s clipboard so they can be paste into a Windows compatible worksheet This is the most common choice Edit gt Clear Results clears the current data from the Results window When selected you are asked if you wish to save the current results A Yes response opens a window to enter the name and path of the data txt file you wish to save File gt Save Results As a txt file with the name and location specified in a Save as window File gt Print Results This opens your computer s printer selection window to set preferences for printing Tip only choose this option if the amount of data isn t excessive since the font typeface will be scaled in size to include all the data on one printed page Proceed to perform an Object Classification of the same image or an Object Analysis of a new image In the former case you will be asked if you wish to save the current object analysis results before they are overwritten by the object classification data In the latter case decide beforehand whether or not to concatenate the object analysis data see 4 2 3 3 Concatenate Object Analysis Results ET olx Note ImageTool minimizes rather than closes the Results window when either the minimize or the X delete button in its upper right corner of its title bar is clicked This wise design prevents unintentional loss
107. ities Small snapshots of these 3 images are provided in Fig in the tutorial worksheet Pause Now that we have covered the preface material we can start this interactive training tutorial First you will set the various preferences to perform CMEIAS object analysis and classification While making these settings please consult the CMEIAS 1 28 User Manual for details describing what each task does and why it s necessary for CMEIAS image analysis IV Pause The Find Objects tab page Setting Preferences Find Objects in the Settings dialog box has settings that relate to the process of finding the foreground objects using the ImageTool brightness threshold procedure For most CMEIAS automated object analyses and morphotype classifications select the Search in AOI select the Automatically Select Objects check Exclude background specify 498 for the Maximum of Objects and set Minimum size pixels of foreground objects to 30 To introduce these settings click OK to display the various preference tabs select the Find Objects tab page enter these settings and then select the Apply and OK pushbutton to activate them for this tutorial Find Objects ADI options Search entire image f Search in ADI Search options C Manually Select Objects Maximum of objects E f Automatically Select Objects T Include objects at edge of image tinimum size pixels 30 IM Exclude background Display op
108. ize while enhancing pixel resolution The next step is crucial in order to adjust the image size to one that would be typically displayed at 1 1 in ImageTool s workspace and analyzed by CMEIAS without creating problems in pixel resolution By default the Resample Image box is checked in the Image Size dialog box Fig 4A with an arrow and the Bicubic resampling algorithm selected Photoshop arbitrarily assigns 72 pixels inch 28 34 pixels cm resolution to all unfamiliar images and then automatically multiplies the image s pixel width and length here 1315 x 1033 pixels by the default resolution of 72 pixels inch to report a print size of 18 264 inches x 14 347 inches If a smaller Print Size dimension e g 5 inches width is entered while the Resample Image box remains checked Photoshop will reduce the Pixel Dimensions by the factor required to maintain pixel resolution at 72 pixels inch This reduction in pixel dimension results in a poorer quality image that cannot be analyzed optimally The solution to this problem is to maintain the pixel dimensions of the entire original image while changing its size Do this by deselecting the Resample Image box setting before reducing the image size Fig 4AB arrow In Fig 4B the image is resized to 5 inches wide automatically the height becomes proportionally resized to 3 928 inches and this automatically increases the pixel resolution to 263 pixels inch to maintain the 1315 x 1033 Pixel
109. lar Rod F Uneaten Dirt a C Ellipsoid em ae Spiral C Prosthecate Done Curved Rod C Rudimentary Branched Rod D U shaped Rod 5 C Branched Filament Cancel Regular Rod DE Fig 30A D The Edit Feature applied to the pseudocolor coded classification image for the CMEIAS Morphotype Classifier 62 5 The Reassign Category Label window will activate and display the checked option button next to the selected object s currently assigned morphotype In Fig 30B this is a coccus 6 Reassign the object s morphotype by clicking its corresponding alternate option button When registered the pseudocolor of the selected blinking object in the Classification result image will automatically change to reflect the morphotype classification reassignment In Fig 30A C arrow the cell has changed from red to blue indicating it has been reassigned as a regular rod Fig 30B 7 Repeat steps 1 5 for any other objects whose morphotype classification needs reassignment 8 After all classification edits are made click the Done command button in the dialog box to accept the editing results Before these are displayed however another window will pop up asking if you wish to save the original unedited data currently in the Results window Most often the appropriate response is No since they contain classification errors that you have identified and corrected A Yes response displays a window to indicate the name and location of the
110. lects the measurement attribute and then enters the upper class limit value for each group or bin class within the range one at a time The number of bin classes up to 16 total will equal the number of upper class limits entered plus one with the last bin grouping objects whose attribute value is greater than the largest upper class limit entered The bin widths range per bin do not have to be constant The full output data from this object classifier are a contingency table listing each bin value range classification data of object counts for each bin class and the mean and std dev of measurement values for all objects in each bin The measurement features added in the CMEIAS code makes this 1 D object classifier adequate for analysis of relative size distributions of all bacteria in the image regardless of their shape and for analysis of microbial communities with relatively low morphological diversity containing only a few morphotypes Any one of the size shape excluding Fourier Descriptors or grayscale brightness measurement attributes can be used see 4 1 1 Measurement Feature Preference Page For instance you can use this object classifier directly to classify the regular rod cocci and filament cell morphotypes in bacterial communities using the Width Length attribute Since the object classification is performed directly a separate prior object analysis is unnecessary The usefulness of this IT CMEIAS 1D object classifier is illustra
111. lls are sampled for community analysis Note The number of images required to capture the entire morphological diversity of the community examined will depend on its morphotype richness distribution of abundance of morphotypes present and the spatial density hence total number of the cells within the images The sample size is adequate when a plot of cumulative morphotype diversity index vs cumulative sample size rises to a plateau asymptote 3 5 Image Acquisition and Sampling Density CMEIAS ImageTool 1 28 performs automatic object analysis and morphotype classification on 8 bit grayscale images without compression acquired either directly via Twain compliant devices digital camera flatbed scanner frame grabber image capture board for live or pre recorded video image or previously acquired digital images saved in one of ImageTool s 22 PC compatible image file formats Tiff BMP etc Do not attempt to perform automatic image analysis in ImageTool using LZP compressed Tiff images BMP Indexed Color images or compressed Jpg images The accuracy of object analysis and morphotype classification strongly depends on the sampling density of the pixels in the digital image defined by their resolution pixels per unit image length and magnification A disadvantage of digital microscopy as compared to photomicrography is that digitization will sample objects at higher error rates because the pixel sampling unit is typically larger in size
112. lution inputs background subtraction interactive histogram image stack averaging etc to process the image before threshold segmentation Consult the UTHSCSA Image Tool operator manual for pertinent details on those image processing routines Examples of interactive editing steps to prepare images for threshold segmentation include 1 Stretch the histogram of grayscale intensity levels to its full 0 255 scale range 2 Apply a median filter to smooth object contours without changing their overall shape or size 3 Increase contrast to help find and separate the foreground objects from background 4 Eliminate invalid pixels whose brightness values lie in the range of foreground objects of interest 5 Fill holes in selected foreground objects with pixels of foreground brightness 6 Introduce a narrow continuous line of pixels with background brightness to split touching objects 7 Add some noise free background pixels to the image margin so the user defined Area of Interest polygon can be easily drawn to fully enclose the foreground objects of interest The ideal combination of image editing procedures will vary depending on the types of segmentation problems encountered e g invalid objects touching objects grayscale background pixels within the brightness range for the objects of interest More details on the editing procedures used to segment images of bacteria are presented in Liu et al 2001 Microbial Ecology 41 173
113. m Ecological Research in Row Crop Agriculture Michigan State Univ East Lansing MI Fukuda M J Matsuyama T Katano S Nakano and F B Dazzo 2006 Assessing primary and bacterial production rates in epilithic biofilms on pebbles in Ishite Stream Japan Microbial Ecology 52 1 9 Dazzo F B 2006 Spatial analysis of Microbial Colonization on Plant Roots Grown in Soil In R Finlay and J Luster eds COST 631 Handbook of Methods in Rhizosphere Research Chapter 4 1 Microbial Growth and Visualization of Bacteria and Fungi Dazzo F B M Schmid and A Hartmann 2006 Immunofluorescence microscopy and fluorescence in situ hybridization combined with CMEIAS and other image analysis tools for soil XXXI 50 51 52 53 54 55 56 57 58 59 60 and plant associated microbial autecology In J L Garland C Hurst D Lipson A Mills L Stezenbach and R C Crawford eds Manual of Environmental Microbiology 3rd ed Chapter 59 pp 712 733 American Society for Microbiology Press Washington DC Dazzo F B and Y G Yanni 2006 The natural Rhizobium cereal crop association as an example of plant bacteria interaction In N Uphoff et al eds Biological Approaches to Sustainable Soil Systems pp 109 127 CRC Press Boca Raton FL Parnell J J Park V Denef T Tsoi S Hashsham J Quesen and J M Tiedje 2006 Coping with polychlorinated biphenyl PCB toxicity physiologi
114. med in on the image to magnify the ends of the bar scale line while still displayed in full view on your monitor screen This is normally done manually to the active image by clicking the zoom in shortcut icon magnifying lens with a sign in the toolbar In the next steps you will spatially calibrate the image for object analysis using the 10 um bar scale located in its lower right corner Click OK to zoom in on the image command zoomin Processing Zoom in command zoom in command zoom in Pause Read the entire text in this message box to spatially calibrate the image First click OK and move the active window of the image containing the objects so its upper left corner lies at the upper left corner of the ImageTool available workspace Then position the cursor at the lower right corner of the image and enlarge it diagonally a couple of inches Then scroll the horizontal and vertical bars to display the entire line of the magnification bar scale located at the lower right corner of the image Next with great care position the tip of the pencil cursor exactly on the left edge of the bar scale and left click once then move the cursor to the same horizontal position exactly on the right edge of the bar scale and double click to create a line of the same length After completing this step a window will display asking How long is the line Check the option button for microns micrometers left click your cursor within the
115. morphotype classifier 6 edit morphotype classification results directly on the image so that the final output of morphotype classification data fully conform to the users specifications and 7 illustrate how CMEIAS morphotype classification data can be evaluated by ecological statistics to compare the morphological diversity richness abundance dominance evenness and similarity of complex microbial communities Pause You should have 2 documents readily available either hard copies or already opened and minimized on your computer while running this training tutorial These are the CMEIAS v 1 28 User Manual Cmeias128 pdf and the CMEIAS v 1 28 Tutorial Worksheet CmeiasTutoral Worksheet pdf Pause The CMEIAS 1 28 Tutorial Worksheet contains 9 tables and 10 figures of data generated by this recorded macro and is provided to enhance the efficiency of this training tutorial plus show various ways to work with CMEIAS ImageTool image analysis data Maximize the tutorial worksheet and take a minute to briefly view the various figures and tables in it now Pause You will analyze 3 different microbial community images that were included in the CMEIAS v 1 28 installation and now are located in your Uthscsa ImageTool Help folder Each image is an 8 bit grayscale binary Tiff containing 170 microbes representing the distribution of size morphotypes and abundance among different bacterial and archaeal populations in methanogenic bioreactor commun
116. n so open the Object Classification tab page deselect that feature then click Apply and OR command preferences Pause Open spatially calibrate and find the foreground objects in the Community A tif image open C program files Uthscsa ImageTool Help Community a tif command zoom in command zoom in command spatial calibrate command zoom out XII command zoom out command find objects Pause For your convenience we have already created a calibration file of the optimized area bin increments for this IT CMEIAS 1D object area classification Click OK to display the window to set preferences for the IT CMEIAS 1D classifier select Area as the measurement attribute from the drop down list box click the Load button and double click the OptimizedAreaBins ocd file to introduce these prerecorded bin increments Then click OK to run the object classification plug in C Program Files Uthscsa ImageTool Plug Ins objclass dll Pause Now repeat these steps to perform a classification of cell areas in the Community B image using the same optimized upper class borders open C program files Uthscsa ImageTool Help Community b tif command zoom in command zoom in command spatial calibrate command zoom out command zoom out command find objects plug in C Program Files Uthscsa ImageTool Plug Ins objclass dll XIII Pause Maximize the tutorial worksheet to view the data on cla
117. of data 50 CHAPTER Performing Object Classification Microscopy commonly reveals various characteristics of bacteria relevant to classification of their morphological diversity in actively growing microbial communities Such diversity can be enormous as illustrated in the Fig 1 micrograph of the bovine rumen microflora which created my spark of inspiration for development of CMEIAS and CMEIAS is designed to extract the information in such community images so that morphological diversity can be quantified In contrast morphological analysis of non growing microbial communities e g unamended bulk soil is much less informative since most of the microbes have differentiated into dwarf nearly spherical quiescent u tramicrocells for starvation survival an ecologically important physiological adaptation during which the cells shut down their cell wall building machinery for their cell division cycle and hence no longer express their distinctive morphotype CMEIAS v1 28 features various measurement attributes and two object classifiers to analyze size and or shape characteristics of microorganisms in segmented digital images of microbial communities and then classifies them into their appropriate morphotype These supervised object classifiers report on the richness of different morphotype classes found within the images and the distribution of abundance among each of them thus providing the ecological data needed to compute various mo
118. of object frequency counts per class and associated basic descriptive statistics To proceed click OK to find and open the image containing the microbes you wish to analyze using this 1 D classifier command open Pause In the Find Objects tab select the parameters you want ImageTool to use to find your microbes of interest Remember to select Search in AOI and Automatically Select Objects if you wish to analyze objects in the image that also contains a bar scale for spatial calibration command preferences XXI Pause In the Object Classification tab select Report on classifications using a single measurement feature IT CMEIAS 1D classifier and the associated parameters to report on this classification Typically these will include the Value Range and Number of Objects optionally you may also include the Mean and Std Dev of all objects in the classification Also select if you want ImageTool to display a new image showing objects pseudocolor coded by their classification command preferences Pause Now perform a spatial calibration of your image if you are using a size attribute e g cell length to classify the objects If classifying by a shape or grayscale level attribute e g width length mode gray level it is not necessary to spatially calibrate the image If the latter is the case click the Escape key on your keyboard to bypass this step or draw a line of any length on the image select pixels and accept i
119. olumn headings for each measurement attribute a column listing all the objects found each numbered the same as in the thresholded image and cases rows of the mean standard deviation and individual values for each measurement attribute extracted from the corresponding objects Normally at this stage you would copy these object analysis data Edit gt Copy Results to a worksheet in your Windows spreadsheet program Edit Paste That s not necessary here since they are already copied to Table 1 of the pdf tutorial worksheet Maximize the tutorial worksheet and view the top portion of that table now Pause Next scroll the tutorial worksheet to view Fig 2 which is a 2D scatter plot of Length vs Width for the cells analyzed in this image Note the locations of clustered objects and outliers Pause View Fig 3 in the tutorial worksheet This is a line scatter plot of the fifth column of Table 1 an ascending sort of each cell s Width Length ratio vs ranked object number plus various grouped labels text horizontal lines and arrows inserted Then minimize the tutorial worksheet and return back to this macro VIII Pause Note the two data points next to small black arrows where obvious breaks occur in the line plot When plotting these data in your spreadsheet application point the cursor over them to display a text box indicating their corresponding X and Y axis positions in the plot the Y value is their computed width length r
120. on 7 22 Steps to Perform a CMEAS 2 Morphotype Classification and Fig 29C 46 6 2 2 Manual Object Counting ImageTool provides a manual object counting feature select Analysis gt Count and Tag or the corresponding Toolbar shortcut for both grayscale and colored images Fig 23 This feature is useful when the goal is to obtain an object count in a grayscale image that cannot be segmented by thresholding or to obtain an object count in a color image When the Count amp Tag feature is selected the mouse cursor becomes a pencil when placed over the active image Point to each object of interest and count it by a single click of the left mouse button This procedure registers a colored dot on the object counted in the image to indicate it has already been counted and the current incremental count in the Count amp Tag dialog box Conclude the counting process by double clicking the final object to be counted You can specify the radius size in pixels and color of the dot register in the Settings gt Preferences gt Count Tag tab page The object count will display in the dialog box and be sent to the Results window as illustrated in Fig 23 Toolbar Analysis Processing Macro Settings Window Help shortcut Paints Ctrl F Count and Tag Ctrl T Manual Count and Tag feature in e F amp ImageTool Ver 1 27 F ar p MIM Ima OVX yen _ x Click on the objects to be counted one at a time
121. on for this thresholding procedure should be gradual for non binary grayscale images so the results can be closely inspected in order to find their optimal threshold setting By following this brightness thresholding procedure one can easily appreciate the importance of acquiring editing images to a high quality so they can be reduced to the foreground objects of interest before object analysis classification The ImageTool download includes the image blobs tif shown below that is useful to develop the skills required to threshold non binary grayscale images so the foreground object contours are found accurately and reproducibly BLOBS ox Ly d S s a L 43 5 8 Manually Select Objects In most cases it is advisable to edit the image sufficiently so that all of the foreground objects of interest can be segmented from background by the threshold procedure using the Automatically select objects feature selected in the Settings gt Preferences gt Find objects tab page 5 7 Find Objects by Brightness Threshold Segmentation However manual selection is a useful alternative to the automatic mode when the image contains invalid objects whose pixel brightness and size still lie within the range that defines the foreground objects This scenario occurs when the image is too complex to easily edit completely and or the AOI polygon cannot easily be drawn to exclude all the invalid objects When Manually select objects feature is
122. on of this Agreement is unlawful void or for any reason unenforceable it shall be deemed severable from and shall in no way affect the validity or enforceability of the remaining provisions of this Agreement This agreement shall be governed by Michigan law UTHSCSA ImageTool License Agreement University of Texas Health Science Center at San Antonio UTHSCSA Texas UTHSCSA ImageTool software both binary and source if released hereafter Software is copyrighted by The Board of Trustees of the University of Texas UT and ownership remains with the UT The UT grants you hereafter Licensee a license to use the Software for academic research and internal business purposes only without a fee Licensee may distribute the binary and source code if released to third parties provided that the copyright notice and this statement appear on all copies and that no charge is associated with such copies Licensee may make derivative works However if Licensee distributes any derivative work based on or derived from the Software then Licensee will 1 notify UTHSCSA regarding its distribution of the derivative work and 2 clearly notify users that such derivative work is a modified version and not the original UTHSCSA ImageTool distributed by the UT Any Licensee wishing to make commercial use of the Software should contact the UT c o UTHSCSA to negotiate an appropriate license for such commercial use Commercial use includes 1 inte
123. otation angle and the deviation of the starting point 25 The corresponding Fourier coefficients are a a T jin2n a a Se Y e n 1 N L If is not an integer number this equation is at approximate equality and the degree of approximation is dependent on the difference between t and its nearest integer number It can be proved using the property of this equation that features f a la n 2 N 1 are invariant with respect to translation scaling and rotation Since low order Fourier coefficients occupy most of the energy of the signal we use fi f fas fos fives fv fy 3 and fys as 8 Fourier descriptor FD features in CMEIAS When the Fourier Descriptors feature is selected in the Preferences gt Measurement Feature tab page Fig 6 data from each object will be reported in 8 columns labeled FDO to FD7 for all 8 Fourier descriptors in the object analysis Results window also see Fig 8 and 4 1 4 Display of Object Analysis Data in the Results Window See studies by Liu et al 2001 for more information on the importance of the Fourier Descriptors in CMEIAS Morphotype Classification Feret Diameter Diameter of a circle having the same area as the object computed as 4Area z_ Future versions of CMEIAS will call this measurement attribute the Equivalent Circular Diameter Centroid x y The center point x y intercept labeled O in Fig 7a is computed as the average of the x y coord
124. ount in the image since that value equals the highest numbered row of data in the object analysis Results window see black arrow pointing to number 70 in the Object array of Fig 24 The only exception to this standard display design of the Results window is in the reporting of Centroid X Y and Gray Centroid X Y object analysis data When selected each of these 2 attributes displays 3 columns of image analysis data for each object found the ImageTool Centroid array contains both the x and y coordinates delimited by a comma in the same worksheet cell and CMEIAS lists these Centroid X and Centroid Y coordinates in separate adjacent cells Fig 8 in pixel units This alternate CMEIAS display facilitates the export and use of these spatial coordinates in other software applications for spatial distribution analysis of objects Similarly the ImageTool GrayCentroid array is followed by the CMEIAS GCentroid X and GCentroid Y arrays Perform a Manual Object Analysis of a linear feature single or multiple segment lines or area of a user drawn polygon within the digital image as follows see image below First specify the measurement data to be collected length area perimeter grayscale luminosity features using the Distance and Area tab pages of the Settings dialog Settings gt Preferences gt Distance or gt Area Then activate the manual object analysis tool Analysis gt Distance or Analysis gt Area
125. ptimized pseudocolor assignments for bin classes 4 ImageTool may crash while thresholding large images containing many objects These problems have been resolved in ImageTool ver 3 0 and CMEIAS ver 3 1 currently under development Let us know via the Register amp Contact Us page in the CMEIAS website lt http cme msu edu cmeias gt of any other problems you encounter using CMEIAS 1 28 in Uthscsa ImageTool 1 28 so we can address them in CMEIAS ver 3 1 Enjoy CMEIAS Frank Dazzo lt cmeiasfd msu edu gt Michigan State University XXVI APPENDIX Vi Studies using CMEIAS References to studies that include data acquired by CMEIAS are posted and updated periodically at the CMEIAS website lt http cme msu edu cmeias gt The list below was updated in March 2010 If you use CMEIAS in your research please send the reference to your published work in the format shown to lt cmeiasfd msu edu gt so it can be added to this list Tiedje J M K Nusslein J Zhou B Xia C Moyer and Frank B Dazzo 1997 The vast world of microbial diversity 5 th JST International Symposium on New Frontiers in Microbiology Tokyo Japan 2 Tiedje J A Fernandez S Hashsham S Dollhopf F Dazzo R Hickey and C Criddle 2000 Stability persistence and resilience in anaerobic reactors a community unveiled Int l Symp on Establishment and Evaluations of Advanced Water Treatment Technology Systems Using Functions of Complex Microbial
126. ptimized upper class limits of 0 0625 and 0 5 Classification count data are reported for each bin class 54 1 Set Settings gt Preferences as follows Find Objects tab page Fig 9 select Search in AOT Automatically select objects Exclude background Show object numbers on original image Show object outlines on original image Object Classification tab page Fig 25 select Report on classification using a single measurement feature Display new image showing objects colored by classification 2 Open the image to be analyzed Image spatial calibration is not needed when classifying objects by shape measurement since they are dimensionless 3 Select Analysis gt Object Analysis gt Find Objects and perform the brightness threshold routine see Find objects by brightness threshold segmentation and Fig 20 to find all the foreground cells of interest in the image There are 170 microbes in the CocciRegularRodFilament tif image used to make Fig 26 and Fig 27 4 Select Analysis gt Object Analysis gt Object Classification to open the IT CMEIAS 1D Define Object Classifications window Fig 26 5 Click the Attribute pull down list box and select the shape measurement attribute desired in the example Width Length is selected Enter the upper class limit for each bin in the Maximum Value in Class input fields 7 Alternatively click the Load button select an appropriate ocd cl
127. r to the left until all foreground objects are separated from background and simultaneously their red filled 42 contour accurately represents their size shape and position as in the original grayscale image Fig 20C After thresholding the image release the mouse button and click OK To threshold negative images see 5 5 Negative Image Transformation slide the left most plunger to the right When done with this interactive thresholding procedure the foreground objects will be numbered and their contours annotated in front of the original grayscale image background as illustrated in Fig 20D Figs 20A D illustrate these threshold segmentation steps for a non binary grayscale image of bacteria requiring no editing of background to fulfill this criterion Fig 20A shows the grayscale image that has been previously edited to split a few touching cells by drawing a white colored line of single pixel width between them arrows using the pencil tool in Adobe Photoshop see 3 7 Image Editing Fig 20B displays the threshold window with the complex histogram of pixel brightness levels in this grayscale image Fig 20C shows the segmented objects colored in red found in this image by adjusting the brightness threshold to eliminate the background pixels Fig 20D shows the resultant thresholded image with each bacterial cell found displayed with an annotated contour and assigned number ordered from the bottom up The interactive slider acti
128. r work Contact us at lt emeiasfd msu edu gt or lt dazzo msu edu gt for feedback or for questions not already answered in our publication on CMEIAS J Liu et al 2001 Microbial Ecology 41 173 194 and 42 215 this CMEIAS 1 28 User Manual our CMEIAS website lt http eme msu edu cmeias gt or the Cmeias128help chm file You must be logged on as an administrator to run this training macro This macro is located in ImageTool Macro Cmeias128training Tutorial itm and you can start it from Macro Cmeias128trainingTutorial The verbatim text of the macro is as follows Pause Welcome to this hands on CMEIAS v 1 28 Training Tutorial which is designed to accelerate your learning of the major important features offered in CMEIAS ImageTool v 1 28 used in the image analysis of microbes The macro was prepared by Frank Dazzo Hassan Hammoud and Jose Zurdo at Michigan State University East Lansing MI 48824 USA For questions or comments on this tutorial macro email F Dazzo at cmeiasfd msu edu or dazzo msu edu Pause Every time this macro advances to the next message box it will automatically run the wav sound file assigned to your Windows Asterisk function You may prefer to unlink any wav file to the Windows Asterisk sound or install one that you would enjoy hearing multiple times during this training session access through sound controls in the Control Panel Utopia Asterisk wav is quite pleasant and is available at http www astrof
129. raham A F Wallis 2004 Quantification of bacterial morphotypes within anaerobic digester granules from transmission electron microraphs using image analysis J Biotechnol Techniques 7 142 148 Chi F S H Shen H P Cheng Y X Jing Y G Yanni and F B Dazzo 2005 Ascending migration of endophytic rhizobia from roots to leaves inside rice plants and assessment of their benefits to the growth physiology of rice Appl Environ Microbiol 71 7271 7278 Ponder M S Gilmour P Bergholz C Mindock R Hollingsworth M Thomashow and J M Tiedje 2005 Characterization of potential stress responses in ancient Siberian permafrost psychroactive bacteria FEMS Microbiology Ecology 53 103 115 Gantner S M Schmid C D rr R Schuhegger A Steidle P Hutzler C Langebartels L Eberl A Hartmann and F B Dazzo 2006 Jn situ quantitation of the spatial scale of calling distances and population density independent N acylhomoserine lactone mediated communication by rhizobacteria colonized on plant roots FEMS Microbiology Ecology 56 188 194 Dazzo F B G Tang G Zhu C Gross D Nasr C Passmore K Kulek E Polone A Squartini A Prabhu C Reddy R Peretz L Gao R Bollempalli D Trione E Marshall J Wang M Li D McGarrell S Gantner J Liu and Y Yanni 2006 CMEIAS v3 0 Advanced image analysis software to strengthen microscopy based approaches for understanding microbial ecology 2006 Annual Mtg Long Ter
130. rdinates for geostatistical analysis of object spatial distribution Some measurement features are abbreviated in the column headings of the Results window These include Max Curv maximum curvature ABR ratio of the object s area to the area of the smallest bounding box WLR ratio of width to length LWR ratio of length to width and FD0 FD7 Fourier Descriptors 0 through 7 These features are defined above in section 4 1 2 97 4 2 Find Objects Settings ImageTool s Find Objects tab page Settings gt Preferences gt Find Objects contains important settings used during threshold segmentation to find the foreground objects in the active image Fig 9 All features in this tab page are controlled by code in Uthscsa ImageTool v 1 27 Wilcox et al 1997 and not by CMEIAS See the It doc operator manual for additional information Find Objects AOI options C Search entire image Search in AOl m Search options C Manually Select Objects Maximum of objects 498 Automatically Select Objects Include objects at edge of image Minimum object size pixels 30 J Exclude background Display options Show object count in a message box Place abject count in results window V Show object numbers on original image Choose Font I Show object outlines on original image Choose Color Fig 9 Find Objects tab page Settings gt Preferences gt Find Objects to specify the settings used during the s
131. re dimensionless normalized measures of cell shape Area Bounding Box Area This dimensionless shape measurement feature is the ratio between the object s area and the area of the smallest rectangle enclosing the object The four boundaries of the minimum enclosing rectangle are parallel to the major axis and minor axis respectively This measurement of shape is approximately computed as Area Major Axis Length x Minor Axis Length Fourier Descriptors Fourier descriptors are shape measurement features derived from the object contour and can be used to represent open or closed curves at different spatial scales In addition shape features can be extracted from Fourier descriptors which are invariant to translation scaling and rotation To compute the Fourier descriptors the object boundary represented as a polygon is resampled by a sequence of equidistant points x y k 0 N 1 where the distance between the neighboring points is a constant Let z x jy k 0 WN 1 be a sequence in the complex space Then z4 can be represented by its discrete Fourier transform coefficients N 1 _jnk2a N Z J ae k 0 N 1 n 0 where N 1 jnk2a a 5 ze N n 0 N 1 N k 0 are the discrete Fourier transform coefficients and ao is the mean of z k 0 N 1 Let z Sz e T k 0 N 1 be a distortion of zx where S is the scaling coefficient T the translation vector in the complex space the r
132. red in micrometers The program draws and labels the bar scale in the lower right corner of the image using the selected foreground and background colors 20 CHAPTER Setting preferences prior to image analysis ImageTool preferences must be set properly before performing an image analysis Access these very important settings by selecting Settings gt Preferences gt various tab pages The most important preferences for image analysis of bacteria using CMEIAS are located on the Find Objects Image Measurement Features and Object Classification tab pages Some of the ImageTool default settings should be retained for CMEIAS operations whereas others need to be changed Consult the ImageTool Ver 1 27 Operator Manual C UTHSCSA ImageTool Help folder it doc for other preference settings not indicated here 4 1 Measurement Features Used in Object Analysis 4 1 1 Measurement features preference page The ImageTool v 1 27 Object Analysis tab page is replaced with a Measurement Feature tab page Settings gt Preferences gt Measurement Features when the CMEIAS v1 28 objanal dll plugin file is installed and run Any combination of measurement features can be selected from this tab page for object analysis The Object Analysis plugin in UTHSCSA ImageTool v 1 27 contains 19 measurement features including object area perimeter Feret diameter major and minor axis lengths roundness elongation compactness major and minor axis angle
133. rphological diversity and community similarity indices of microbial communities The Object Classification tab page Settings gt Preferences gt Object Classification used to select the two CMEIAS ImageTool v1 28 object classifiers and associated features is illustrated in Fig 25 Options include selection for the type of Object Classifier report on objects individually see details on this feature in the ImageTool 1 27 It doc operator manual not very useful here for CMEIAS v 1 28 and display a new image with objects pseudocolor coded according to their assigned classification 51 Object Classification Attributes to send to results window f Report on classifications using a single measurement feature IY Value range classification M Mean value of all objects in class IM Number of objects in class IM Std dev of all objects in class f Report on CMEIAS Morphotype Classifier using multiple measurement features Report on objects IY Display new image showing objects colored by classification Fig 25 The Object Classification tab page Settings gt Preferences gt Object Classification to select the features that activate the CMEIAS ImageTool v 1 28 object classifiers 7 1 ImageTool CMEIAS 1D Object Classifier Developers of ImageTool designed the first object classifier to sort objects based on division of a scale defined by a single measurement feature To run this one dimensional classifier the user se
134. ry with clean tissue paper 4 Place slides horizontally and spaced apart on a perfectly leveled surface that is checked with a bubble spirit level Note lab benchtops may have areas that are not perfectly level 5 Dispense 1 6 ml of the dissolved tempered agarose solution from a pipette in a zigzag motion over the clean smooth surface of each slide frosted end side up without allowing overflow 6 Cover without touching the slides with a large inverted glass dish until the agarose solidifies 7 Dry the agarose coated slides overnight in a horizontal position at 50 C in a desiccator oven 8 Store the dried agarose coated slides in a clean slide box until used 3 3 Preparation of Dispersed Microbial Samples 1 Label the frosted end of the slide with the sample name s 2 Pass the suspended sample of the microbial community rapidly through a 25 gauge needle several times to assist in achieving a uniformity of single cell dispersion 3 Dilute the dispersed sample until it is visibly barely turbid 10 cells ml 4 Immediately before microscopy deposit exactly 26 ul of the sample to a confined area of the dried agarose coated slide This volume is optimized for the next step adjust if using a different coverslip size Each slide can accommodate two samples 5 Carefully apply a cleaned 22 x 22 mm coverslip to the suspended sample without trapping air bubbles The coverslips should have the proper thickness matched to your 100
135. s displayed in the Results window should match Table 2 in the worksheet Pause Our next task is to perform a CMEIAS comparative analysis of the microbes in the Community A and Community B images illustrated in Fig 1 of the tutorial worksheet We will first analyze their cell size distribution using object area measurements and the IT CMEIAS 1D object classifier Click OK to open the Community A tif image open C program files Uthscsa ImageTool Help Community a tif Pause Now spatially calibrate this image using the 10 um bar scale using the same steps described earlier Click OK to perform the spatial calibration command zoom in command zoom in command spatial calibrate command zoom out command zoom out Pause Now find the 170 foreground objects using the same threshold steps described earlier making sure to exclude the bar scale from the AOI polygon that you draw on the image command find objects Pause Click OK to open the Settings dialog box select the Measurement Features tab page and select the Area attribute for object analysis deselect all other choices followed by Apply and OK command preferences Pause Now click OK to perform the object analysis of cell areas plug in C Program Files Uthscsa ImageTool Plug Ins objanal dll Pause Next click OK to open the Community B tif image spatially calibrate it using the 10 um bar scale and perform the threshold routine to find the 170 for
136. se the polygon automatically by double clicking to connect a final straight line from the current position of the cursor to the start position The polygon line will be blue when located in the image background Fig 18 and will be yellow when it covers an object To include objects of interest at the edge of the image add a little extra white margin to the image canvas beforehand e g in Adobe Photoshop select white background then Image gt Canvas Size so sufficient space is available to draw the polygon to enclose all of these foreground objects of interest The image will then display annotations on the objects Fig 19 see 4 2 3 4 Show Object Numbers on Original Image and 4 2 3 6 Show Object Outlines on Original Image and also a message box indicating the total number of objects found if this feature is selected in the Find Objects tab page Figs 9 and 19 see 4 2 3 1 Show Object Count in a Message Box If Find Objects gt Search Entire Image is selected instead of Search in AOI recommended when no bar scale is present the thresholding procedure will find and annotate the objects in the binary image without the need to draw the polygon enclosure nra eT gm e a 65 ImageTool SA G There are 70 objects in the image Fig 19 Find objects in a binary image Shown is a binary image with each object annotated by an ascending consecutive number and a colored contour plus the ImageTool information window indicating the total number o
137. sing tools before it is closed Color images are saved as 8 bit indexed color files with the same pixel dimensions This feature is unchecked by default and isn t commonly used during CMEIAS work sessions 4 3 3 Show full path names in image title bars This ImageTool feature displays the full path and filename of the image beginning at the drive letter in the title bar of the image window If not checked then only the filename will be displayed Choose the latter when working with images that are too narrow to display the full filename 4 3 4 Initial zoom scale factor These option buttons are used to select the zoom factor when opening a new image Select 1 1 unless you optimize otherwise For example open small 128x128 images at a zoom factor of 2 1 but select 1 2 or lower to scale down images that are wider or taller than your screen so they can display fully within the ImageTool workspace This selection does not affect the zoom commands you can still zoom in and out of the image as normal it just affects the initial display of the images when opened 34 CHAPTER Image display adjustments calibration and thresholding 5 1 Load Image Both grayscale and color images can be opened and analyzed manually in ImageTool but the image must be an 8 bit uncompressed grayscale for automatic object analysis and object classification Load an image in ImageTool by one of 3 ways 1 click File gt Open 2 click on the Op
138. ssification computations statistical analysis and interpretation of data aA BWN eR 3 1 Image Requirements for Image Analysis 1 The first requirement for accurate image analysis is to produce a very high quality primary image using any type of microscopy e g brightfield LM phase contrast LM TEM SEM CLSM that can distinguish the contour of each foreground microbial cell of interest from the background Low quality images are the most frequent cause of unexpected inaccurate and unreliable results using CMEIAS For CMEIAS image analysis the primary image can be acquired in photo video or digital format but it must be converted to an 8 bit digital grayscale image in an uncompressed file format eg Tiff in order to be analyzed and classified automatically in Uthscsa ImageTool CMEIAS Ver 1 28 RGB color images can be opened and analyzed manually but not automatically in ImageTool for example see 6 2 2 Manual Object Counting Digital images must have adequate magnification and pixel resolution so that even the smallest cell of interest can be sampled with sufficient pixel density at least 30 pixels object to define its contour for accurate morphotype classification This requirement poses no problem for skilled microscopists using research quality optics and a modern computer Prior to analysis the image must be edited sufficiently so that it can be reduced to the foreground objects of interest using the brightness
139. ssification of object areas for these two community images in Table 4 and the corresponding results plotted as vertical sequential stacked and vertical clustered bar graphs in Figs 6A and 6B These plots illustrate another way that CMEIAS IT can extract quantitative data indicating the similarities and differences in distribution of microbial abundance within different communities Pause Now we are ready to introduce the CMEIAS morphotype classifier which is the most innovative and powerful feature of CMEIAS v 1 28 This object classifier uses various pattern recognition algorithms optimized for 11 major microbial morphotypes represented by 98 of the genera described in the 9th Edition of Bergey s Manual of Determinative Bacteriology Scroll to Fig 7 in the tutorial worksheet to view the hierarchical outline of characteristics for these 11 morphotypes classified by CMEIAS including each one s specific pseudocolor coded assignment Pause In order to perform these automated morphological classifications using CMEIAS v 1 28 we must first measure all required shape attributes of the microbes in an object analysis of the image These measurement features are defined mathematically in this CMEIAS 1 28 User Manual the Cmeias128help chm file and our Liu et al 2001 Microbial Ecology 41 173 194 publication Click OK to open the preference window select the Measurement Feature tab page click all 7 shape measurement features located in the left framed
140. t anaerobic bioreactor communities community a tif and community b tif Each image contains 170 microbial cells is included in the download of Cmeias v 128 and is used with the CMEIAS v 1 28 Training Tutorial Appendix I Figure 32A is a 2 D Ranked Abundance clustered bar plot of the morphological diversity in these 2 communities Figure 32B is a 2 d cumulative sequential vertical stacked column plot indicating the proportional distribution of cell area as a measure of cell size for each morphotype in these 2 communities Tables 2 and 3 report the results of an ecological statistics analysis of the diversity indices and community similarity coefficients computed from these morphotype classification data using EcoStat Software Howard Towner Trinity Software lt http www trinitysoftware com lifesci5 ecostat7 html gt For other examples of CMEIAS data analysis e g community ecological succession see Liu et al 2001 Microbial Ecology 41 173 194 and the Morphotype Classification and Publications using CMEIAS pages at the CMEIAS website http cme msu edu cmeias Also the itm training tutorial and its accompanying worksheet pdf file provided in the CMEIAS v 1 28 installation includes steps to follow the generation of typical data graphics and computations of CMEIAS community analysis 66 Fig 32 A A vertical cluster bar graph of the ranked abundance of microbial morphotypes in two communities and B a cumulative vertical st
141. t the morphotype assignment if necessary To proceed click OK to open the image containing the microbes you wish to classify command open Pause In the Find Objects tab select the parameters you want ImageTool to use to find your microbes of interest Remember to select Search in AOI and Automatically Select Objects if you wish to analyze objects in an image that also contains a bar scale for spatial calibration command preferences XXIII Pause Object analysis data must first be extracted from the objects in order to classify their morphology In the Measurement Features tab select all 7 check boxes for the shape measurement attributes within the framed box labeled CMEIAS Morphotype Classifier These 7 check boxes actually represent 14 different shape measurement features since 8 Fourier Descriptors are grouped into one check box You may also select any other measurement attributes that you want to include in the object analysis step that must precede morphotype classification Then click OK to register these preferences command preferences Pause In the Object Classification tab select Report on CMEIAS Morphotype Classifier using multiple measurement features Also select Display new image showing objects colored by classification so you can inspect the morphotype classification results Then click OK to register these preferences command preferences Pause The image does not have to be spatially calibrated
142. ted in Figs 26 and 27 where cells in the image are classified into bins according to their optimized size area and shape width length ratio attributes respectively The CMEIAS 1 28 training tutorial also includes a section of image analysis using this IT CMEIAS 1D object classifier The CMEIAS v 1 28 download includes a CmeiasCalibrations folder containing 20 customized ocd files with various bin widths that can be used with the Load function to help optimize the range of upper class limits used with this object classifier 52 7 1 1 Using the IT CMEIAS 1D Object Classifier for Cell Size Classification 1 Fig 26 Set Settings gt Preferences as follows Find Objects tab page Fig 9 select Search in AOT Automatically select objects Exclude background Show object numbers on original image Show object outlines on original image Object Classification tab page Fig 25 select Report on classification using a single measurement feature Display new image showing objects colored by classification A ji mye NP Fig 26 Size classification of bacteria using the CMEIAS ImageTool 1 Object Classifier with optimized area bins Shown are the original image the selection page of measurement feature area and optimized upper class limits the pseudocolored classification image and the output of classification data in the Results Define Object Classifications x Attribut
143. ting technology to process color images for quantitative microbial ecology studies at single cell resolution Microbial Ecology DOI10 1007 s00248 009 9616 7 2010 Microbial Ecology journal publication 59 2 400 414 3 D Wilcox B Dove D McDavid and D Greer 1997 UTHSCSA ImageTool Version 1 27 Operator Manual Univ Texas Health Science Center at San Antonio 59 p Download website http ddsdx uthscsa edu dig itdesc html 4 J Russ 2002 Image Processing Handbook 4 Edition CRC Press Boca Raton FL http www reindeergraphics com CMEIAS Software is copyrighted by Michigan State University Direct any questions regarding CMEIAS not covered by this help document or ref 1 3 to Frank Dazzo at lt emeiasfd msu edu gt Credits amp Acknowledgments Major support for development of CMEIAS has been provided by the Michigan State University Center for Microbial Ecology with funding from the National Science Foundation the US Egypt Science amp Technology Joint Program MSU Research Excellence Funds Center for Microbial Ecology Center for Renewable Organic Products Center for Microbial Pathogenesis MSU Kellogg Biological Station Long Term Ecological Research Project and the Michigan Agricultural Experiment Station We thank Jim Tiedje Rawle Hollingsworth Brent Dove Don Wilcox Phil Robertson Martha Mulks Stan Flegler and the CMEIAS software development team of microbiologists mathematicians and computer scientists listed below
144. tion 12 ccc1ec1e00 53 7 1 2 Using the IT CMEIAS 1D Object Classifier for cell shape classification 2 1 11 00 54 7 2 CMEIAS 2 Morphotype Classifier cccccece eee ee ee eee eee e eee e esate eee eae ea eea eee eeee 55 7 2 1 Hierarchy of characteristics for CMEIAS microbial morphotype classification 57 7 2 2 Steps to perform a CMEIAS Morphotype Classification ccccccccecsecsececeenecneneceeseeeesenans 58 7 2 3 Important points regarding the CMEIAS Morphotype ClasSifier 2 2 cccccceceeeneeseneeeeeeee 59 7 2 4 Sources of morphotype classification error and the CMEIAS edit feature 2s s10000 61 7 2 5 Editing the type 1 misclassification error in a CMEIAS Morphotype Classification 62 7 2 6 Editing the Type 2 unrecognized class and Type 3 invalid object classification errors 63 8 GRAPHICS and ECOLOGICAL STATISTICS of CMEIAS OBJECT ANALYSIS amp CLASSIFICATION DAT AY ti sccscncesacns cxseckentepzcincaiesenxenen aE AR aE EDERE 66 APPENDIX CMEIAS 1 28 Training Tutorial Macro ccecceeeeeeeeeeeeeeeeeeeeeeeeeeeeeeaeees APPENDIX II CMEIAS Object Analysis Macro cccceeeeeeeeeeeeeeeeeeeeseeseeeeeeeeeaeeees XIX APPENDIX IIl ImageTool CMEIAS 1D Object Classification Macro 2 eseseeeee XXI APPENDIX IV CMEIAS 2 Morphotype Classification Macro ccccccceeeeeeeeeeeeeaeees XXIll APPENDIX V Known Problems in UTHSCSA Im
145. tions Show object count in a message box 7 Place object count in results window IY Show object numbers on original image Choose Font IY Show object outlines on original image Choose Color command preferences Settings Preferences Pause Next under the Display options in the Find Objects tab page select Show object count in a message box Show object numbers on original image and Show object outlines on original image Then double click within the Choose Font and Choose Color rectangular boxes to specify your preferred choices followed by the OK pushbutton We recommend that these latter two image annotations be set as a blue 14 pt Tahoma font and a bright magenta color click Color to display the palette select the rightmost cell in the 2nd row to define the object contour Click OK to return to the Find Objects tab page and then introduce these settings followed by Apply and OK command preferences Pause Next click OK to open the Settings Preference Image tab page and set the initial zoom ratio to 1 2 for this tutorial followed by Apply and OKR command preferences Pause Now open the Measurement Feature tab page Settings Preferences Measurement Features in the Settings dialog box to specify the CMEIAS ImageTool measurement attributes to be extracted from each microbe in the Ist object analysis The User Manual and Help file describe the formulas used to comput
146. tructions accordingly when using other operating systems Software Requirements e CMEIAS v 1 28 upgrade of Uthscsa ImageTool Ver 1 28 e Image editing software e g Adobe Photoshop Image Processing Tool Kit CMEIAS Color Segmentation e A document display program to read print the CMEIAS pdf files e g Adobe Reader e A spreadsheet program to paste analyze and save CMEIAS data e g Microsoft Excel e Ecological statistics software to compute community diversity e g Trinity EcoStat 2 2 Download amp installation of ImageTool CMEIAS ver 1 28 An easy to use Cmeias128setup exe is now available for free download at the CMEIAS website lt http eme msu edu cmeias gt to install the CMEIAS ver 1 28 upgrade of the UTHSCSA ImageTool host program core files plus numerous other new and revised user support files of CMEIAS To install CMEIAS ImageTool double click the setup wizard file within Windows Explorer accept the UTHSCSA and CMEIAS license agreements and answer the prompts when indicated The installer places a CMEIAS IT 1 28 shortcut icon on your desktop and a CMEIAS start menu that allows direct access to launch the image analysis program chm help file pdf user manual pdf tutorial worksheet for the training macro and link to the CMEIAS website Fig 2 If ImageTool v 1 27 is already installed on your computer the setup wizard will replace its original executable and utility library core files with the CMEIAS v 1
147. ts default dimension command spatial calibrate Pause Next select the thresholding method e g Manual or None image is already thresholded and perform the brightness threshold procedure to find your foreground objects of interest command find objects Pause Next select the measurement attribute for this classification from the drop down list box Then in the Maximum Value in Class input fields enter the upper class limit for each bin class you wish to include in the object classification The bin widths that define the dimension of each class do not have to be equal Alternatively you may load a previously saved ocd file to enter the desired series of upper class limits in the input fields of the maximum values in each class CMEIAS v 1 28 contains numerous ocd calibration files that can be used with the Load function of this IT CMEIAS 1 object classifier To use a previously saved calibration file click the Load button select the desired ocd file located in the Calibration folder and then click Open The upper limit for each class should display in the Maximum value in Class input fields After introducing these parameters click the OK button to perform the object classification on the thresholded image plug in C Program Files UTHSCSA ImageTool Plug Ins objclass dll Pause Your object classification data should now be displayed in the Results window You may Print or Save these data using the File main menu
148. tures from each object in the image The computing time required for this step will vary depending upon the speed of the computer the number of thresholded objects in the image and the combination of measurement features selected The computing time required to analyze the shape attributes of the objects in the binary image used to make Fig 24 took between 1 2 seconds on a Pentium III PC running at 700 MHz Fig 24 Extraction and display of object analysis data in the Results window Note that the object count in the image is indicated by the highest numbered row of object analysis data arrow in the Results window corresponding to the object found at the highest position in the image arrow 48 Following automatic object analysis each selected measurement feature appears as a column heading in the Results window spreadsheet Fig 24 also see Fig 8 and 4 1 4 Display of Object Analysis Data in the Results Window and the corresponding values extracted from each foreground object found in the image are reported individually as cases rows in units that are designated during the calibrate spatial measurement step see 5 6 Spatially Calibrate the Image The mean and standard deviation for all measured values in each column array are automatically computed and displayed in gray filled cells of the first 2 rows below the column headings Figs 8 and 24 This object analysis routine is the 5 way 4 automatic to obtain the object c
149. unity images included in the CMEIAS 1 28 download They are fully segmented tiff 8 bit grayscale images edited to binary They each contain 170 bacteria representing the distribution of size and abundance among several different populations distinguished by their morphological diversity in anaerobic bioreactor communities We suggest you open and view these images in ImageTool before starting the CMEIAS training tutorial so you can fully comprehend the important point that microscopy can reveal significant morphological diversity in complex actively growing microbial communities This tutorial will illustrate the ability of CMEIAS v1 28 to unleash its awesome computing power by quantifying the similarities and differences in morphological diversity of complex microbial communities Every time this macro advances to the next dialog box it will automatically activate the wav file assigned to your Windows Asterisk function for Win 2000 Start gt Settings gt Control Panel gt Sounds amp Multimedia gt Sounds gt Sound Events gt Windows gt Asterisk gt OK for WinXP Start gt Control Panel gt Sounds Speech and Audio Devices gt Sounds and Audio Devices gt Sounds gt Windows gt Asterisk gt OK Depending on your preference either unlink any wav file to the Windows Asterisk sound or install whatever wav file you would enjoy hearing multiple times during the training session Utopia _Asterisk wav is quite pleasant and is avail
150. ve these data using the File main menu or Clear Cut or Copy these data to the Windows clipboard using the Edit main menu plug in C Program Files Uthscsa ImageTool Plug Ins objanal dll APPENDIX Hil ImageTool CMEIAS 1D Object Classification Macro CmeiasIT 1objectClassification itm If this is the first time you are using CMEIAS we recommend that you perform the exercises in the CMEIAS 1 28 Training Tutorial Macro Appendix I beforehand using the images provided in the program download and installation You must have administrator rights to run this macro file If you want to view the data while they are being extracted from objects in your image optional recommended then before starting this macro you should maximize the Results window position its worksheet near the right edge of the ImageTool workspace and adjust its size to display 3 4 columns and expanded to the full viewable height of your graphical user interface Pause You can use this macro to help guide you through the steps used to perform a 1 dimensional object classification on your own images using the IT CMEIAS 1D classifier This analytical tool is designed to classify objects found in an image based on any single measurement attribute except Fourier Descriptors Centroid x y and Gray Centroid x y featured in CMEIAS ImageTool v1 28 and up to 16 classes created by the corresponding upper class borders entered by the user The classification output data consist
151. white input field and use your keyboard to enter a length of 10 00 Then click OK command spatial calibrate Settings Calibrate Spatial Measurements command zoom out Processing Zoom out command zoom out VII command zoom out Pause The next step is to find the foreground objects of interest to be analyzed in the segmented image using the ImageTool brightness thresholding procedure Pause In this thresholding segmentation step you are asked to select the threshold method to find the foreground objects of interest For this example select None followed by OK since the image contains only black and white pixels binary Then use the mouse cursor to draw a blue line polygon on the image click once at each corner to enclose all foreground objects while excluding the bar scale in the lower right corner To close the polygon double click the last corner Once completed each foreground object found within the polygon will automatically become surrounded by a colored line and numbered consecutively from the bottom up command find objects Settings Preferences Find Objects Pause Click OK to perform an object analysis on the foreground microbes found in the image plug in C ProgramFiles Uthscsa ImageTool Plug Ins objanal dll Analysis Object Analysis Object Analysis Pause Take a look at the layout of your object analysis data reported in the ImageTool Results window It lists a row of c
152. window option is selected Therefore do not select this feature when collecting concatenated object analysis data from multiple images in the same dataset described next Furthermore object analysis data in the Results window always include a column of object numbers Figs 8 amp 10 so the object count in the image can always be obtained from that data list 4 2 3 3 Concatenate Object Analysis Results In general the morphological diversity of a complex microbial community cannot be fully represented by a single microscopical image When building an object analysis dataset from multiple images in the same community CMEIAS ImageTool v 1 28 can directly concatenate the data in the Results window for each selected measurement feature for all valid objects up to the 498 object limit in all images constituting the same dataset when the feature Place object count in Results window in the Find Objects tab page is deselected Fig 11 and the Results window is not cleared between object analysis cycles In this case the rows of numbered objects analyzed for multiple images are concatenated Fig 11 i e the data extracted from each new consecutive image will restart the numbering of objects in the Results window as 1 The computed mean and standard deviation of all the measured values within each column of the Results window will automatically update as the new object analysis data of the most recently analyzed image are concatenated
153. with the upper class limits values The various ocd files provided in the CMEIAS 1 28 installation can be used with the Load feature to help optimize the upper bin limits for this classifier 7 Verify that the Attribute selection hasn t been changed occurs when loading a calibration file made with another attribute and then click OK to run the object classification 8 The pseudocolor classification image and classification data will display onscreen Fig 26 7 1 2 Using the IT CMEIAS 1D Object Classifier for Cell Shape Classification Cell shape classification using the IT CMEIAS 1D classifier involves the same 7 steps as listed above 7 1 1 Using the IT CMEIAS 1D Object Classifier for Cell Size Classification except that the image does not have to be spatially calibrated step 2 since shape attributes are dimensionless and a single shape measurement feature is selected step 5 When run the pseudocolored classification image and the classification data will display onscreen Fig 27 Define Object Classes Results Attribute width Lenath Value Range Std Dev Classifications Masimum Yalue in Class 1 0 0625 2 jos zl aoi Fig 27 Cell shape classification using a single measurement feature and the IT CMEIAS 1D classifier The example shown is a morphotype classification of cocci blue regular rods green and unbranched filaments red using the Width Length Ratio measurement feature and o
154. y edited images with an overall accuracy of 97 The major source of this 3 error rate is the occasional presence of cells whose morphology lies within the real world continuum that overlaps the assigned borders of closely related microbial morphotypes that are defined in 14 dimensional space by the pattern recognition algorithms in the CMEIAS program Consult our major publication on the CMEIAS morphotype classifier Liu et al 2001 Microbial Ecology 41 173 194 the CMEIAS 1 28 User Manual or the Cmeias128help chm file for a complete discussion of the sources of morphotype classification errors and how CMEIAS is designed to minimize and address them Pause We will now illustrate this type of error and how to correct it in a CMEIAS morphotype classification using the Community B tif image Pause Find the short thick red pseudocolored cell near the center of the classification result image a white arrow points to it in the worksheet Fig 9A Azotobacter chrococcum has this morphology CMEIAS classified this coccobacillus as a coccus rather than a regular rod because of its high roundness and circularity shape values and its length is less than twice its width However some CMEIAS users may prefer to classify it as a short plump regular rod instead CMEIAS was designed to facilitate the users desire to edit such borderline object classifications within the pseudocolor classified image and revise the object classification data accordingly
155. ype evenness All parameters are computed in EcoStat Trinity Software from inputs of CMEIAS data on the number of different 68 morphotypes substitute for species found and the distribution of abundance among each class The indices of community characteristics include measures of morphotype richness diversity dominance and evenness These community similarity analyses compare the similarity and dissimilarity of morphological diversity between the two communities Some parameters are significantly influenced by sample size and the 2 community images analyzed here each contain equal number of cells Consult pg 15 16 of Howard Towner s EcoStat pdf manual for mathematical details These features illustrate how CMEIAS can strengthen the microscopy based approaches that compliment other methods e g 16S rDNA based nutritional versatility FAME etc of polyphasic analysis to characterize the structure and function of complex actively growing microbial communities in situ without cultivation Final reminder images to be analyzed by CMEIAS must be of high quality and 8 bit grayscale and each foreground object of interest must contain at least 30 pixels and have a brightness range allowing it to be segmented from all background pixels and be found using the thresholding routine in ImageTool Accuracy depends foremost on the quality of the primary image Enjoy CMEIAS Frank B Dazzo Michigan State Univ 69 APPENDIX CMEIAS

CMEIAS Ver. 1.28 - The Center for Microbial Ecology

Contents

Download Pdf Manuals

Related Search

Related Contents