Home

DIGIXCT SOFTWARE USER MANUAL

1. 51 3 3 2 Optional Geometry calibration 55 3 8 3 Mandatory Geometry definition 62 3 4 Ao AA AA AA AA 64 3 4 1 Mandatory Input definition 64 3 4 2 Optional Projection pre processing 67 3 4 3 Optional Phase contrast pre processing 72 3 4 4 Optional Projection correction 73 3 5 X ray CT system compatibility 76 DigiR3D is compatible with the following CT system Mand AG UN EE AR 76 3 6 Mandatory Output definition 77 3 6 1 Volume format and name 77 3 6 2 Volume dimensions to be reconstructed 78 3 6 3 Region of interest 80 3 6 4 Volume resolution 82 3 6 5 Cylindrical reconstruction constraint 82 3 6 6 Reconstruction grid orientation 84 Our Reconstruction mandatory 84 3 7 1 Filtered back projection 85 3 7 2 lterative algorithm cannes msmmsumnitees 87 3 7 3 IPOSDPDIOGSSSIDIO entente restent 93 3 8 Preview and Adjustment 97 3 8 1 Preview Settings 97 gt PCS a a ceseasessenteassaeaneockes 101 3 8 3 Metal artifact reduction 104 3 9 Save load and reconstruction 106 4 ANALYSIS OF CUTTING PLANES DI
2. 9 1 1 HUQ CC 0 eR ARA APR NAT 9 1 2 Minimum configuration requirements 9 1 3 Tusj ee 10 I AA 10 1 4 Protection system nenne 10 1 5 Diagnostic window 11 VOLUME VISUALIZATION DIGIOBS MODULE 11 2 1 Presentation of the default GUI 12 2 2 Management of preferences 12 2 2 1 Computation AD rs 14 2 2 2 Interface tab 15 22 3 NN 18 2 3 Loading a volume 19 2 4 Volume visualization eese 21 2 4 1 Multi display visualization 21 2 4 2 Volume slab visualization 22 2 4 3 Display parameters 23 2 4 4 The color palette 26 2 4 5 Light sources a 29 246 OPjeCtrerleCliOM 2 hi 30 2 5 Managing a volume 31 2 5 1 Display management 31 IG 2 5 2 Volume alignment Xa 33 2 6 Data management 36 2 6 1 Saving and exporting volumes 36 2 6 2 Exporting slices 3 2 6 3 Exporting videos 39 264 SONOS OUS a ee mn ANE 40 TOMOGRAPHIC RECONSTRUCTION DIGIR3D PLUG IN 50 3 1 MUOU 11e EE 50 3 2 Workflow definition ccccccseeceeeeeeeeeceseeeeeeseeeeeeaees 50 3 3 Hardware dellhillOftessaoszubi cse deniers S a sies Ras 51 3 3 1 Mandatory Camera definition
3. R FS KO ev a ALPA BALO 4 v NES p 4 L d L CAE Y PES KA VAS Y c SJ PRU SI INR Pay cum BAN D v Apo rab MAX AGA Aa 7 Se D FRA TAT IN En va Figure 5 9 wireframe display of our example 5 2 3 4 Advanced parameters Advanced parameters propose the choice between two mesh extraction algorithms Dual Contouring OrHierarchical algorithm Hierarchical approach is an optimization of Dual Contouring In some cases it can reduce the number of triangles in a mesh rA A ACT AST SORS ae a away a C I LOT e N Vi N vi VIN VUVVN gt VV AT Y VAY AC YA KW v NINA KANA RAD AVAVA OC VW AA v AA BAS RANEA w ws a NATAVAN aA RS ITS LOE OTA PC I A PAVAN AT A E DIR IS LM S TNO NIT TLR Ki 137 qm fA DigiXCT crane 256V2 File Window Tools Help CEEE He a N v F ZVV v IN KA 4 NN VV V VA AAA 7 AAA VIA VIA Va BESTS NAA V VA AN Wi X Y 7 UN LA KA NA AT V VE D AP VA JA IV D 4 L x A Ses ASS RISK AVA VAN AND e EL vas CLS ARS paaa rae PIN pg M wy AAS KIA Var ar Ararat KA AV AVA T vi o tale t SES aes EET d Surface Surface 1044956 triangles V Advanced parameters Method Dual Contouring 4 Roughly 786432 Subsamplng Auto t Denst
4. Slice As lt lt Prev Next gt gt Help Shift F1 Save image Ctrl S Level curve mode Ctrl L k Pa v Measure mode Ctrl M uli Merge measures FI Save measures e Clear all Point S Segment Lo Angle Pa Circle radius imago sre zum age TI biko 4 Circle border rx rn aes JJ 4 Auto anne DE e Draw profile Figure 4 6 Different measures proposed 4 2 4 2 1 Measure reference All necessary information about reference can be found in chapter 2 5 2 4 2 4 2 2 Point To select a point in the image you need to e Select point measure e Left click on desired pixel in the image 111 Displayed information is e X Y Z location in measure reference e Corresponding grey level in the pixel Cut plane Cut Radial Plane Cut Cyinder Cut Cone ok Omm Parameter 04 Omm 18 Omm Thickness 0 027 mm Reset Stack As Sice As cPrev Apply Nexto Resolution Contrast Zoom Image size white 3 436 t r z gamma 15 pb black 0 708 eJ LJ Pixel size 0 0266833 mm Volume contrast B 1v Figure 4 4 Points measure 4 2 4 2 3 Length measure To measure a length two methods are available e First method o Select segment measure N 112 o Verify that the assisting mode is deactivated l o Left click with the mouse to select the origin point of the segment o Left click with the mouse to select the ending point of the segment
5. 1 3 Installation The installation of DigiXCT begins from the execution of the file DigiXCT X XX X exe where X XX X corresponds to the installer version tag Once the installer has been started let yourself be guided by the installation wizard For any difficulties in installing or launching the application please contact Digisens technical support 1 4 Protection system The DigiXCT software suite is protected by a license file and a USB key which is programmed and supplied by Digisens It must be inserted before starting DigiXCT and must remain so when it is running otherwise the application will shut down The key is installed when it is used for the first time this only needs to be confirmed by clicking on the Next key when the Windows dialogue box Hardware device detected appears Once the installation is finished you must click on Finish An information bubble appears by the menu bar Start Windows with the message Your hardware is installed and ready to use 11 1 5 Diagnostic window When the application is launched if the graphic configuration is not adapted a diagnostic window appears Figure 1 1 It shows graphics card reference OpenGL version that is present and OpenGL extensions necessary which have been found or not When the software is used for the first time you can choose to disable this information window for future start ups In the event of display issues this information should be given to the t
6. 89 Digisens product also implements more standard methods based on Algorithm ART Algebraic Reconstruction Techniques This algebraic method is used in reconstruction of tomography data The algorithm implements the formula With VE MR M v Vig V A l prm current estimation of the volume Af Matrix of projection of the volume on the plan of the image sensor Ri and factor convergence A ART formula is interpreted as follows the error between the measures and the projection of the estimated volume is back projected on the object space An update of the volume is obtained by averaging the corrections of a set of projections subset If this set contains only one projection then the algorithm is a SART algorithm If it contains all projections then the algorithm is a SIRT algorithm The algorithm OSSART is a variant consisting in the launch of several SIRT with subsets containing a set of images lower than the total number of projections The iteration number must be fixed by the user as the low frequencies are reconstructed first it is necessary to increase the iteration number to reconstruct the details of the object The convergence parameter convergence speed must be in the interval 0 2 It balances the correction which is applied to the current volume It can be chosen manually or automatically it depends of the subset size Auto mode 90 Figure 3 45 Example of the reconstruction
7. Figure 3 42 Ring artifact reduction Left image raw image with ring artifacts Right image filtered image Important note Ring artifact filter works only if the object is in its natural orientation vertical axis by default merged with the axis of rotation If the 95 grid orientation is used to rotate the object after the reconstruction this filtering is not functional The user will be warned of it by a message in the screen If however the rotation is compulsory it will be possible after reconstruction to rotate the object by using the reference axis modification described in the chapter 2 5 2 Important note Ring artifact filter is not available in slice preview mode 3 7 3 2 3D edge preserving filter This filter is a 3D extension of the filter described in 3 4 2 3 3 7 3 3 Bilateral filter Bilateral filter is an edge preserving noise reducer In homogeneous regions a Gaussian kernel is used to smooth pixel intensity in a range defined by kernel size Weighting function is composed of two Gaussian kernel A first kernel based on distance from processed pixel defined by its standard deviation Oxyz A second kernel based on pixel intensity describing intensity gap defining edges The kernel is defined by its standard deviation o Figure 3 43 shows how the filter works left image shows the signal to process middle image the kernel shape and right image the resulting processed signal To use this filter see figure 3
8. licic MUISGISENS DIGIXCT SOFTWARE USER MANUAL DigiXCT v3 0 RevA January 2012 Revision history Modifications January 2012 For any technical question please contact Digisens at Digisens SA 17 all e du lac d Aiguebelette BP278 Savoie Technolac 73375 LE BOURGET DU LAC Cedex France Phone 33 0 479 658 917 Fax 33 0 479 250 899 Email support digisens fr General information Copyright ODIGISENS S A 2012 All rights reserved The reproduction transfer distribution or storage of part of or all of the contents of this document in any form whatsoever is forbidden without the prior written permission of DIGISENS All the other products and company names mentioned in this document may be the trademarks or brand names of their respective holders The software and plans presented in this document are the property of their respective holders DIGISENS reserves the right to make changes and improvements to all the products mentioned in this document without prior notification Users must periodically check by using appropriate comparative methods the conformity of their software If this is not done DIGISENS cannot be held responsible Under no circumstances can DIGISENS be held responsible for loss of data for financial losses or any other sort of special or accidental damages be they directly or indirectly caused in any way whatsoever 1 2 Contents GENERAL PRESENTATION OF DIGIXCT
9. the data is granted if its value is 50 2 times more importance with the data that with the regularization Positive constraint the option Constraint to positive values fixes all negative value in the reconstruction to 0 values Volume nriors Regularization Method Strength 22 Constraint to positive values Figure 3 41 Regularization 3 3 Post processing You can apply to the reconstructed volume 3D post processing to improve your image quality uH UISISETIS 94 3 7 3 1 Ring artifacts remover Ring artifacts have several origins All of them are connected to an incoherence of pixels behavior with radiation We can quote for example dead pixels or pixels with the variability of their response to illumination with regard to the others In this last case the incoherence between the flat field image and the acquisition can be at the origin of ring artifact in the reconstruction Several approaches exist to correct these pixels Several tools are proposed in pre processing of the projections or the sinogram itself The tool proposed here is applied to the reconstructed volume The principle is based on the detection of rings and their deletion Both accessible parameters in expert mode are the maximal thickness of rings to be deleted size and Alpha which takes into account if rings are complete and constant or modulated ie partial rings By default parameters value is 3 for size and 0 2 for Alpha
10. 14 012 mm 15 376 mm 2 Cylindrical constraint Resolution vowel A 513 vox Y 503 vor Z 503 vox gt Best 50 um 100 um 5 mm Other Memory used 495 1 MB voxel size 23 5828 um x 27 8569 um x 30 5686 um in 80 chunks File Window Tools Help w e BERTA KP EF Z Plane ective 1000 Figure 3 34 Definition of the reconstruction area of interest 82 3 6 4 Volume resolution The resolution of the volume for each of its dimensions is shown figure 3 35 You can also directly set the resolution of the reconstruction either by clicking on the buttons corresponding to a pre defined resolution value or by choosing Other which allows a value to be put in whatever the resolution Once the resolution has been chosen object dimensions in voxels are updated The resolution of the volume on each of the dimensions is indicated see figure 3 35 So reconstructed voxels are not necessarily of the same size on the three axis Resolution voxel X 923 vox Y 750 vox Z 750 vox Be Mum 50mm O0w oo imm 5mm fier Memory used 1 9 GB voxel size 207 444 um x 207 444 um x 207 444 um in 288 chunks Figure 3 35 Volume resolution 3 6 5 Cylindrical reconstruction constraint The oylindrical constraint allows limiting the reconstruction to a region of the space corresponding to the intersection of the back projection of all the available projections This intersection is centered on t
11. Optional Projection correction During a tomography scan the flux emitted by the X ray tube varies generator variations generator or tube regulations The consequence is contrast variations in acquired images As long as it is possible to monitor these variations it is possible to correct them To process image sequence check the box Flux Variation see figure 3 27 This processing need a reference image used as a flux reference for all other images in the sequence Ideally this image should be a white image image acquired without object and with the same conditions To display this image check the box display reference image P kumanan UTSISC1 IS Figure 3 26 Top image corresponds to a reconstructed slice of a carbon plate with phase effects Bottom image corresponds to the same reconstructed slice with BAC approach 19 Then define the ROI for measurement by modifying the red square extensions Just modify its width and height and locate it in the region you need to monitor V Flux variations etere C DATA XCT ref_image tif p Xmin 0 i Xmax 959 i display reference image Ymin 0 Ymax 767 Figure 3 27 Flux variation correction parameters After X ray emission begins the target of the tube is warming Because of this temperature increase the X ray spot drifts leading to projection shifts The consequence on high resolution reconstruction is straightforward and significant for scans longe
12. 44 check the bilateral checkbox and fix the two filter parameters one for intensity and one for spatial characteristics Figure 3 43 Kernel definition a ee Note that those parameters are not absolute values meaning that S corresponds to intensity standard deviation relative to the min max range of grey levels in the reconstruction Its max value is consequently 1 This is the same for S but regarding volume resolution Bilateral filter S1 xyz 014 5g 0 1 Figure 3 44 Bilateral filter parameters The implementation of this filter is fast but the speed is in straight relation with parameters value Figure 3 45 Effects of bilateral filter reconstructed slice of a granite sample 3 7 3 4 Sharpness filter In tomography application system resolution is usually degraded by focus spot finite size detector pixels and scintillator reconstruction algorithm This degradation can be seen as a blurring kernel making reconstruction edges smoothed after reconstruction process To avoid such degradation Digisens has implemented a filter able to retrieve sharp edges lise seu iw 97 To use this filter see figure 3 46 check the sharpness filter checkbox and fix the three filter parameters Number of iterations used for processing A radius corresponding to a low pass filtering necessary to make filtering stable A speed parameter corresponding to a discretization step If the value is decreased filter
13. Segment limits are represented with small squares e Second method o Select segment measure u o Select the assisting mode o Left click in the neighborhood of the desired origin point of the segment o Left click in the neighborhood of the desired ending point of the segment The assisting mode integrates an edge detection tool based on the analysis of grey levels along the profile drawn by the segment Segment extremities are then adjusted to fit the real borders of the part one wants to measure Les extr mit s du segment s ajustent alors au plus pr s du contour de l objet affinant ainsi la mesure Segment limits are represented with small vertical bars Displayed information is e Segment length 113 Cut plane Cut Radial Plane Cut Cylinder Cut Cone ok Omm Parameter O je Omm 18 Omm Thickness 0 027 mm Reset Stack As Sice As Prev Apply Next gt gt gt A LZ ZI E Resolution Contrast Zoom Image size white 3 436 x 424 pix black 0 708 LJ Pixel size a Volume contrast 1 1lv Y Figure 4 5 Segment length measure 4 2 4 2 4 Angle measure To select an angle in the image you need to Select angle measure 77 Left click to select the first point of angle definition Left click to select the second point of angle definition Left click to select the third point of angle definition Displayed information is 114 e Angle value be
14. a given place in the palette figure 2 16 then select the command add from the menu which appears To choose the color used at this grip you must double click on the grip and then select a color from the palette Figure 2 17 It is possible to slide horizontally the grips using Alt left mouse button Transparency adjustment is done directly using the little dot or the triangle placed on the grip and using Ctrl left mouse button Vertical movement upwards increases opacity and downwards vertical movement increases transparency Display is updated automatically The Apply button allows validation of the palette and the min max ranges chosen on the histogram lt triggers the update of data in the 3D visualization window Figure 2 15 Visualization of the 3D part with pre defined display colors Grey Table Heat Table Rainbow Table and Blue Red Table l dicisens 28 Id HNESN 0 525745 Hans Display range selection min 0 151762 S max 0 525745 V Show display range Color presets z EER 10900000 Add grip Er EX Hes Figure 2 16 Density profile Wl DigiXCT turbine 256 bh30 LE File Window Tools Help qw fs IL ER Perspective uM Define Custom Colors gt gt mecum T O Display rang
15. base origin Base When parameters are fixed click on Apply button to get the corresponding slice Note regarding object dimensions the computation may takes several seconds Buttons lt lt Back and Next gt gt makes the cutting plane moves along the normal axes direction with a moving step equal to Thickness Views are automatically recomputed Button Stack As grabs a set of images with cone based surfaces which base radius ranges from Rdown Rdown RSpan Each image corresponds to a cone base radius increased by a radius step equal to Thickness Button S1ice As grabs only the current displayed slice For a truncated cone the radius Rtop is equal to 0 Then the height is computed in such a way the slope of each cone is parallel The figure 4 12 depicts the two cone types used for extraction The user can set the origin of the image first pixel by fixing the Start Angle value This value acts like an offset from 0 to 360 degrees along the perimeter of the base The Reset button relocates the cylinder axis and base to initial values Note in this mode the image size is automatically computed and is not editable 124 Rtop Rdown Rdown Figure 4 12 Stack slice representation in cut cone mode 4 3 4 Cut Radial Plane This mode is an option of DigiCUT Please contact Digisens for further information When Cut Radial tab is selected Figure 4 13 a transparent frame and its axis of revolution are displayed i
16. given in the console Window gt Console Thus you obtain a surface corresponding to the density grey level of the histogram of 0 024 in the 3D display By default the surface and the volume are superimposed The option Hide allows you to hide the voxel model m The information on the right of the name of the surface indicates the number of triangles used to define the mesh In this example we get 786 432 triangles Surface 1 Advanced parameters The name of a surface can be changes at Roughly 786432 any time Subsampling Auto l dicisens 130 Figure 5 3 Mesh extraction from the human head using an iso value of 0 2 5 2 3 2 Post processing Post processing Post Processing such as smoothing and or triangle decimation can be applied to extracted surface Figure 5 4 Post Processing B X m digisens 131 1044956 tangles 7 Advanced parameters Subsampling Auto Roughly 786432 triangles Dry 028 Compute Decimation 10449562 unges m1 100 triangles Optimisation V Foldings angle threshold 120 8 V Small volume threshold v Remeshing Figure 5 4 Post Processing window digisens 132 5 2 3 2 1 Smoothing Smoothing Two parameters allow the user to adjust Region size 30 mesh smoothing smoothing Region Sharp edge impact 22 Size and Sharp edge Impact Smoothing allows the user to re arrange triangles in the mesh defining the surface Region Size is th
17. gt m Measure Position Resolution Contrast Image size white 3 436 Mm 424 pix gamma 154 424 pix black 0 708 Pixel size 00266833mm vo re contest Figure 4 9 Measure example 119 4 3 Specific slice extraction modules DigiCUT plugin manages with several extraction types planes or curvilinear surfaces cylinders or cones 4 3 1 Cut Plane Cut Plane tab computes a 2D slice image from 3D object using cutting planes The Apply button computes the current cutting plane To locate the cutting plane the user can fix the values of the plane normal and plane center filling Normal and Center boxes The user can also move the clipping plane in the 3D view using the mouse in clipping plane management mode Buttons Back and Next makes the cutting plane moves along the normal axes direction with a moving step equal to Thickness Views are automatically recomputed Button Stack As grabs a set of images from the top of the volume to its end along the normal axis direction Note The step between two slices is fixed by the value Thickness Button Slice As grabs only the current displayed slice 4 3 2 Cut Cylinder This mode is an option of DigiCUT Please contact Digisens for further information When Cut Cylinder tab is select Figure 4 10 a transparent cylinder and its axis of revolution are displayed in the 3D view window This superimposed object displays the curvilinear surface used to pe
18. mean value of the entire image stack If this measure is impossible just let a O value in the field set white level e For some systems the acquisition software generates automatically attenuation images meaning that the correction from flat field image the offset image and the computation of the log is already done If this is your case just click on Attenuation images radio button Once the images have been loaded you must indicate to the software the angular range This can be done by filling the field Angles definition in DigiR3D can handle full scan angular range of 360 degrees or partial scans acquisitions at least 180 degrees beam divergence This field can be edited meaning that you can write any value in it Angles definition Angularrange 360 You have the opportunity to make some additional correction on images When the check box Expert user IS checked you can compensate images from electron noise offset by selecting the corresponding image see figure 3 19 Use defect map correction consists in eliminating pixels or pixel clusters with abnormal response it means with null or saturating response whatever signal received In tomography field of application this correction is mandatory to remove the well known ring artifact The image used for this processing has to be binary white level for abnormal pixels and black level elsewhere 67 Image masking Use offset Use defect map 4 Expert user ple
19. modifying the threshold value will paint pixels associated to high density value in red see figure 3 54 Preview settings Parameter adjustment Metal artefact reduction J Use metal artefact reduction threshold gi Compute slice Figure 3 53 Metal artifact reduction GUI Z me coule bis I 0 2 105 Bosco AT TC EN gt File Window Tools Help So ERARSLE Figure 3 54 High density material segmentation When the segmentation of high density pixels is good enough for the i J user compute a slice preview using or a 3D preview using n to check the result If threshold value is validated launch the reconstruction if not refine the threshold Figure 3 55 shows the reconstruction after metal artifact compensation pp MG UIS ISU IS 106 Figure 3 55 Reconstructed slices left image without MAR correction right image with MAR correction 3 9 Save load and reconstruction The load and save buttons ed KA allow the reading and the writing of all of the reconstruction parameters in a calibration file in a proprietary format of which the extension is cal When you click on the reconstruction button t a dialogue box summarizes the defined parameters for a final validation with the effective launch of the calculations ma babai iniba LID 107 4 Analysis of cutting planes DigiCUT plug in As with all the plug ins in the DigiXCT software suite DigiCUT is protected by a software key 4
20. object distance SDD is the focal distance and Umax is the detector width Bottom illustration shows the shifted detector 55 3 3 2 Optional Geometry calibration DigiXCT provides a calibration tool to determine required parameters to achieve a good tomographic reconstruction rotation center focal length based on the capture of some radiographic acquisitions from a calibration test pattern This test pattern is provided by Digisens To get more information please contact our sales team This feature is protected by the license 3 3 2 1 Calibration procedure A tomography scan including calibration step has to take place as follows uone o 103130 1 X ray acquisition of an object is carried out in the tomography system The scan consists in the acquisition of a predefined number of images under different angles of view 2 Atest pattern of size compatible with the previously scanned object has to be acquired with the same system set up fast recording is carried out with a complete rotation and a reduced number of images a minimum of 60 projections and always a multiple of 4 3 The calibration plug in estimates the geometry of the system based on the processing of previously acquired calibration dataset At the end of the process the user exports calibration results in a cal file 4 Then the software is able to 56 generate a 3D reconstruction of the scanned objet ready for vi
21. processing must be started on an image where the spots are completely on the left or on the right of the field of view v The speed is not regular an image must have jumped or been processed twice The calibration sensing procedure must be re visited in such a way as to have regularly spaced out images v A spot collides with the edge and the automatic search detects a sharp variation in spot diameter The parameters uMin uMax vMin Or vMax must be used to trim the images or this spot should not be selected for the calculation v Ina particular angle or in a certain part of the sensor the probe support is too contrasted which prevents the spots from being properly detected The contrast setting must be re visited v The most horizontal trajectories tend to introduce greater centering errors The one or two spot s nearest the center of the image should be deselected v A spot can shift in the probe during sensing In this case the trajectory will be badly centered It should not be selected in the probe When the calculation is successfully completed calibration tool posts a results table see figure 3 11 and spot trajectory is displayed see figure 3 12 WI C DATA XCT Delpix 20110706 2011 07 06 15 19 19 mire demo ima0001 tif Calibration results Probe 7 spots 66 275mm Image center x y 151 812 139 079 0 029427 mm x y 412 532 377 931 0 0799646 pixel Rotation axis x y 2
22. released NITRO algorithm a new iterative based reconstruction approach 3 7 2 1 NITROQO reconstruction method Digisens developed NITRO method for the following purposes e to provide the user with an easy to use tool e to provide improved image quality with few noisy projections see figure 3 44 or from strongly degraded acquisition set up e to allow reconstruction of region of interest see figures 3 45 NITROO is an iterative approach tuning automatically its initialization its convergence speed and its regularization strength The aim is to provide the best image quality whatever scan conditions Non expert user will just need to select NITROG method and to launch the reconstruction For expert one the tab Volume Processing Will provide an access to the following parameters e Regularization from TV approach Volume nriors v Regularization Method vo Strength 75 DE Constraint to positive values Figure 3 41 TV regularization parameters 88 e Regularization from the number of iterations Volume oriors Regularization Method ITERATIONS terations Constraint to positive values Figure 3 42 Regularization from the number of projections e Positive constraint Volume priors Regularization Method ITERATIONS Iterations 206 Constraint to positive values Figure 3 43 Positive constraint Figure 3 44 Example of the reconstruction of a mouse from 24 noisy projections 3 7 2 2 AHT based family
23. resolution Hence filtering projections is a balance between signal to noise ratio and resolution That is why the reconstruction module proposes several apodisation filters each one making its own balance ramp filter for high resolution reconstruction but with lower signal to noise ratio than a combination of a ramp filter and a Welch filter that increases signal to noise ratio but reducing resolution The figures below show the shape of combined filters in Fourier space Hence it can be seen that spare filter ramp filter without apodisation restore resolution regardless to signal to noise ratio On the opposite i 86 Welch filter combined with the ramp filter increases signal to noise ratio regardless to resolution Using those apodisation filters leads the user to accept a trade off between noise level and resolution square bartlett welch hann hamming Figure 3 40 Examples of apodisation effects regarding noise level and reconstruction resolution 87 3 7 2 lterative algorithm Despite its ease of use and speed analytical approaches suffer from strong artifacts in case of degraded projections low contrast low SNR and or degraded scans limited view angle limited number of projections Iterative algorithms help in retrieving better image quality in such contexts Unfortunately such approaches are not as easy as FBP FDK to tune That is why Digisens
24. surface Post Processing and display properties Display properties Some computations on surfaces may generate objects containing hundreds millions triangles This triangle amount depends on the chosen iso value part complexity and mesh resolution In order to generate a reasonable amount of triangles the software proposes a subsampling step increasing the dimensions of an elementary triangle This parameter is of course informative but highly recommended because if the necessary amount of RAM needed to compute such mesh is over the limit fixed by the operating system an application crash may occur and the user can adjust it digisens 128 The button Compute computes the corresponding mesh 5 2 3 Surface operations Here we provide details of the accessible functionalities for each surface created tab corresponds to each surface created Figure 5 3 5 2 3 1 Surface computation In the example shown in 5 2 the interesting density ranges are those between 0 163 and 0 638 Figure 5 2 Volume visualization of a human head To extract a surface from this volume you must 1 Create a new surface New button 129 2 In this tab called surface 1 by default enter the density value corresponding to the surface to be reconstructed iso value at 0 200930 in our example 3 Click on the Compute button to generate the surface The Estimate button helps to define the density value The result of this estimation is
25. value increment e Init value e Init value increment Resulting slices are displayed in the main window 5 103 Note for the first step the initial value used corresponds to the one set up in the geometry definition tab Figure 3 52 Examples of Z slices corresponding to the center of rotation positions 0 05 0 0 0 0 0 et 0 05 0 0 from left to right When processing is complete to inform which new value to take one has to click on the radio button corresponding to the best reconstructed slice The aim is to set the best image in the center of the different views and then click on Accept geometry button to update the geometry dialog tab 104 3 8 3 Metal artifact reduction Metal artifact is a well known artifact in reconstruction of complex parts with very high disparity in density level This artifact can degrade reconstruction in such a way that no analysis is possible see figure 3 53 information signal around the higher density material is degraded by streak artifact or a 0 density area information lost This artifact has two origins e Beam hardening see chapter 3 4 2 4 e Photon starvation DigiR3D proposes to the user a tool dedicated to the restoration of information in the area surrounding the high density area To do so please follow the steps Check Use metal artifact reduction checkbox Select an initial threshold value gt gt 1 Click on Compute slice to compute a slice preview Than
26. 0 00751367 0 999972 0 0 000 deg found clockwise CW direction Rotation Center Src Rot 129 86 0 48 mm Rot Cam 184 47 1 61 mm Src Camera 314 33 113 mm Zoom 242049 0 00023 Figure 3 11 Calibration result example 62 The program cannot find a solution with less than 3 spots In practice at least 5 or 6 are necessary to have error calculation results Ten or so covering the whole of the area to be reconstructed would seem to be optimal tes Kag ESANG i 2 pa Ka i E ce NN ZA Figure 3 12 Spots trajectory 3 3 3 Mandatory Geometry definition The tab Geometry describes the geometry of the system used to acquire the dataset The following figures depict the definition of geometry parameters HEIGHT 63 Geometry Ass degrees Center X Omm Y 0mm z Omm Source Position X Omm Y Omm Z 400mm Ref Point Upper Left x 50mm 8 Y 50mm Z 100mm Figure 3 15 Reference point definition 64 DigiR3D handles several beam profiles You can select them from the combo box figure 3 16 Source type Source geometry Cone beam Figure 3 16 Beam profile selection Specific case of fan beam reconstruction Select a dataset in multipage raw format this means a single file containing the sinogram in the following arrangement widthx1xheight Define raw file in DigiR3D GUI Select PrismY geometry this means dive
27. 1 General presentation of DigiCUT The Cutting plane Analyzer command icon in the plug ins button bar or Tools gt Cutting plane analyzer in the menu allow a 2D cutting plane to be visualized and analyzed Figure 4 4 Figure 4 4 Cutting plane analysis dialogue box 108 4 2 DigiCUT functionalities 4 2 1 Resolution Image size and pixel size are set into Resolution area It allows the modification of pixel dimensions in millimeter and image resolution in pixels 4 22 Contrast Contrast can be set using white black gamma buttons Volume contrast button fixes the image contrast using 3D volume histogram range 4 2 3 Moving in the image The Zoom zone is composed of 9 buttons v Button Fit Fixes the size of the image to the window size v Button 1 1 Fixes the size to the scale of 1 1 1 image pixel corresponds to 1 screen pixel v Button Centers the image in the window v 4arrows Moves the image in the window v Button Zoom It is also possible to zoom in on a precise rectangle by holding down the shift key and then clicking the left mouse button by moving then releasing the button 4 2 4 Other features The help button opens the help window for the different commands as well as the short cuts The Save image button allows the current image to be saved The close button closes the window A right click on the slice image allows a contextual menu to appear The menu has th
28. 3D volume visualization to be compiled When DigiSIM is loaded by DigiCT the La icon appears in the plug ins icon bar 6 2 DigiSIM functionalities To launch the DigiSIM plug in you need to launch the command on the File gt Import Slices menu or click on the m icon to make the import dialogue box appear Figure 5 1 The Input slices zone allows slice files to be specified the gt button displays a dialogue box to select one of the slice files The list of slice files is then displayed in the entry box on the left of this button and a window displays the first image from this list From then on it is possible to scroll the slices via the cursor Figure 5 1 It is possible to select an area within the images by specifying umin umax vmin and vmax and a special group of slices by specifying wmin and wmax Input resolution then allows the resolution of the input images to be entered 141 Em digisens ec oo m Slice scrolling cursor Pl lt gt a herd ber KO Pio oO e rm Figure 6 1 Slice import dialogue box 142 The Contrast zone allows the level of black and white to be modified black and white labels or the images to be inversed via the Invert button The Output volume zone allows the name of the output file to be specified the 3 button displays a dialogue box to choose the placement of the volume and its name An existing file can be overwritten by selecting the header file
29. 5 Reset Stack As Slice As lt lt Prev Apply _ Next gt gt Resolution Contrast Zoom Image size white 3 436 x r gamma 16 424 pix black 0 708 amp SJ LJ Pixel size 0 0266833 mm Yume contrast _ 1 1 v Figure 4 7 Circle measure 4 2 4 2 6 Profile measure To measure a profile along a segment you need to e Select segment measure e Select profile edition e Measure the segment The following figure illustrates the profile edition 117 Cut plane Cut Radial Plane Cut Cylinder Cut Cone 06 Omm Parameter 18 Omm 0 Omm Thickness 0 027 mm Va 1 SAN Resolution Contrast Zoom Image size white 3 436 gt r 7 gamma 18 x 428 pix black 0 708 5 LIP Pixel size Figure 4 8 Profile measure 4 2 4 2 7 Other functionalities Properties edition of the measure is made by clicking with the right button on the concerned point and selecting Properties The dialogue represented on figure 4 9 appears It is possible to modify the color of the measure Color the size of the text Text Size the number of decimals Decimals widths of the lines of measure Width 118 Right clicking on a point also allows you to remove it or to remove the measure Cut plane Cut Radial Plane Cut Cylinder Cut Cone 0 i2 Omm lal Parameter 05 0 mm 14 0 mm Thickness 0 027 mm Reset Stack As Slice As lt lt Prev Next gt
30. D volume The user can also change the size of icons by selecting Icon size among the 4 available formats m DigiOBS global preferences Computation interface Graphics Measure units Selected color SSSS J Outine color J Icon size Small v Multi view configuration 7 Split window at startup Display plane names Sagittal Transversal Coronal 7 Draw lines in single view v Automatically dock window at startup oa J cms Figure 2 4 Preferences window interface tab The Multi display configuration allows the parameterization of multi display as will be described in chapter 2 4 The first box allows the default activation upon launching DigiObs of a multi view work window The Display plane names box allows the X Y Z slices plane to be renamed in the GUI as Sagittal Transversal and Coronal views By default these names are Front Side and Top Draw lines in single display allows the tracing of the edges of the planes intersecting the active display in the visualization Finally the box automatically dock window at startup allows the analysis window to be integrated into the main DigiXCT window as shown in figure 2 5 pr UIS SCI IS 17 File Window Tools Help e cem wEeem Perspective 1641903 Display range selection min 0 937712 S max 3 436278 8 Show display range Color presets BE P7 10000000 WP WP WP Wy ww 4 Add grip Almeno R
31. E 1208231 1 E LI p 21 n YI v LU A y i I i t 4 3 i 3 3 Display range selection min 0 151762 3 max 0 525745 gt Show display range in oe Color presets cy BEBIIG000000 Ped Add gip Add smooth grip Add hard grp HE migi ens a Figure 2 12 Parameterization window choice of the density zone to be displayed turbine 256 bh30 LE File Window Tools Help HE amp amp 0 Perspective Ti g 4 t Display range selection mn 0 151762 gt max 0525745 gt 4 Show display range Color presets EE BII060606000 Add gip Add smooth gip Add hard grip Eg om Figure 2 13 Focus in on the density range A contextual menu in the histogram window allows data to be exported in 26 Excel format right click on the histogram zone then select Export data 2 4 4 The color palette The setting of color parameters is carried out with the palette located under the density profile in the window Volume Display Parameters Figure 2 14 There are 5 pre defined palettes Panel table presets The color palettes can be recorded and loaded via the Load and Save buttons Figure 2 15 shows palette application on a specific example Figure 2 14 Pre defined display colors It is possible to add key points known as grip To do so you must right click the mouse in
32. GICUT PLUG IN 107 4 1 General presentation of DigiCUT 107 4 2 DigiCUT functionalities esssssesssssssss 108 4 2 1 ROSONI e c 108 4 2 2 Contrast nennen 108 4 2 3 Moving in the image 108 4 2 4 Other features nenne 108 4 3 Specific slice extraction modules 119 4 3 1 COR ES a E a i 119 4 3 2 Cut Cylinder aaa 119 RS KULONG eo eee ee ene er ee eee 122 4 3 4 Cut Radial Plane 124 8 5 VISUALIZATION AND EXTRACTION OF SURFACES DIGISRF PLUG IN aanak ANNA cxtat 126 5 1 General presentation of DigiSRF 126 5 2 DigiSRF functionalities 126 5 2 1 SUNACEMANAQE 2 vsert 126 5 2 2 Surface operations 127 5 2 3 Surface operations 128 5 2 4 Simultaneous visualization of multiple surfaces 138 5 3 Saving reloading 139 6 IMPORT OF TOMOGRAPHY SLICES DIGISIM PLUG IN 140 6 1 General presentation of DigiSIM 140 6 2 DigiSIM functionalities eeeesseessssssss 140 7 AUTOMATION OF RECONSTRUCTION TASKS BATCH 7 1 General presentation 143 T2 Batch mode functioning 143 1 General presentation of DigiXCT 1 1 Introduction DigiXCT is a software solution belonging to Digisens Company wh
33. able color matrix B W and slow disable Fragment Program fast Figure 2 6 Preferences window graphics tab The actions of each of the boxes see figure 2 6 are as follows e FAILSAFE opaque volume allows all of the openGL options to be invalidated The final rendering is very poor It is however 19 compatible with any type of graphics card e Disable volume render deactivation of 3D rendering mode by ray tracing method This mode is only recommended for graphics cards which have performance equal to or better than NVIDIA 8800 8600 cards e Fragment Program shared texture and palette extension palette texture extension pixel map color table color matrix texture 3D increasingly slower and decreasingly powerful display mode e Force max texture size to current graphics cards know how to manage 256 or 512 depending on the type of card 2 3 Loading a volume This action is possible either by using the command Open Volume of the menu File or by dragging down a file vol Dragging down a file consists of left clicking on the file holding down mouse button and then dragging the cursor into the application window then releasing the button Figure 2 7 The loaded volume file s name appears in the window title Figure 2 8 To manipulate the volume you only need to use the mouse and the keyboard NOTE For an optimal navigation in DigiOBS GUI it is strongly advised to use a mouse with a roller wheel l
34. adha DE o En x NA DigiXCT alu300 LI File Window Tools Help SRE E Il RT Perspective 1511326 E 3 43628 x Ek Display range selection min 0 937712 max 3 436278 gt Show display range Color presets diia BE BI1I060606000 Add grip Add smooth grip Add hard grp Bo Ba om Figure 2 5 Top image histogram window not docked with the interface Bottom image window docked with the interface Default configuration corresponds to docked windows To get back free displaying of the windows the user can either tick the box lt 18 Automatically dock window at startup in the preferences or left click on the window banner and still keeping the button pressed down drag the window to the middle of the main DigiObs window To dock a window you only have to double click on the upper bar of the window or drag it to the edge of the main window 2 2 3 Graphics tab DigiOBS offers a powerful and realistic 3D rendering tool To enable this visualization module you only have to tick the volume render feature as described below m DigiOBS global preferences Computation Interface Graphics J Force max texture size to V Volume render enable lighting on cutting plane in volume render mode Warming V e render paramete E Advanced openGL options disable Textures 3D FAILSAFE opaque volume B W not editable disable pixel map slow disable color table slow F dis
35. an individual compression of the images is required or not For custom slices the user will use DigiCUT plugin The slice export is carried out with the visualization parameters previously chosen in 3D in gray levels In the export directory concerned the program automatically records an html file which pulls together in a table all of the data concerning the volume and the tomographic slices See below an example of contents generated during the export of 1024 slices 1024x1024 pixels in the direction of the X axis 39 X axis export number of slices 1024 slices image resolution 1024 x 1024 pixels image size 2cmx2cm pixel size 19 53 um x 19 53 um Volume information Resolution 1024 x 1024 x 1024 voxels Size width 20 mm height 20 mm depth 10 mm Voxel size 19 53 x 19 53 x 19 53 um Expanded space 41GB required memory Compressed space required disk Oe basename C Reconstruction Object vol slices per 65 65 65 65 65 65 65 65 65 65 65 sub grid 65 65 65 65 65 60 Display range 0 0 95 Disk storage in 17 subgrid s It is possible to make a video which allows the slices to be displayed one after the other To do this you only have to tick the Create movies for the selected slices 2 6 3 Exporting videos The command Window gt Export Movie allows a video to be exported in the 3D visualization zone either in AVI format save images box not ticked or in the form of a series of images see f
36. as well as the file format jpg png eps ps ppm bmp and fig This function can also be activated via the button E gi digisens 41 Example 1 Filament light bulb 42 Example 2 Molar mUISISENS 43 Ti tps ng Example 3 Hornet gi dicisens 44 l JET L TE amp 4 E t Kan yA i Example 4 Bone 45 Example 5 Distribution pump 2 E OISISETIS 46 Example 6 Casing OISISENS Example 7 Talisman museum 47 ser Heu 0 2 48 Example 8 Human face and skull uH UTSIS EI 49 Example 9 Combination of human face and its skull 50 3 Tomographic Reconstruction DigiR3D Plug in 3 1 Introduction DigiR3D is the tomography reconstruction module of the DigiXCT software suite It produces the cartography of the attenuation of an object based on a stack of X rays projections acquired from several angles of view DigiR3D commands are accessible in the Tools gt 3D Reconstruction Parameters menu or the toolbar Figure 3 1 This command opens the acquisition geometry configuration and reconstruction parameters dialogue box and launches the reconstruction of the volume lu Figure 3 1 The FDK Configuration icon on the toolbar 3 2 Workflow definition Figure 3 2 shows the DigiR3D GUI docked into the main DigiObs window As can be seen DigiR3D consists of 5 toolboxes 1 Hardware definition 2 Input 3 Output 4 R
37. ase refer to manual for details on advanced parameters Figure 3 19 Additional image correction Important note Projections displayed are as they will be used in reconstruction step This means that projections background has to be dark and the part brighter 3 4 2 Optional Projection pre processing Several filters are available to pre process projections before using them in the reconstruction process Pre processing for reconstruction Median filter Kemel size Gaussian denoising filter Kemel size Edge preserving denoising iteration number Beam hardening Coefficient Figure 3 20 Image preprocessing list 3 4 2 1 Median filter Median filter is applied on projections It is often used for noise reduction on images This filter is well known for its performances as salt and pepper noise remover In tomography it will be very useful to eliminate the dead pixels always saturated or always switched off or pixels with a singular behavior Their effect will be the introduction of ring artifacts 68 The median filtering consists in replacing a pixel by the median value of a sorted list of the nearby pixels value The sorting is made from the smaller to the higher value The number of pixels of its neighborhood is defined by default in a value of 1 pixel on both sides of the pixel to be corrected 3x3 kernel The size of the kernel can be increased to 5x5 maximum The choice of kernel size requires a particular attention on
38. behalf of the user As this filter replaces the current value by the value of one of the neighboring pixel it introduces a distortion in the image As a consequence a filter with a too large kernel size will tend to degrade strongly the resulting image The figure 3 21 illustrates this 3 4 2 2 Gaussian filter Gaussian filter is applied on projections It is often used to reduce noise in images The principle of this filler is to average the pixel with its neighborhood The weighting applied to every neighbor takes into account the distance of the neighboring pixel to the processed pixel This filter removes the noise but ads blur to the corrected image The number of pixels constituting the neighborhood is defined by default in a value of 1 pixels on both sides of the pixel to be corrected filter kernel of size 3x3 In expert mode the kernel size can be adjusted The figure 3 22 compares the various techniques of filtering 69 Pa EMEN ut C df Figure 3 21 Median filtering with variable kernel size on image containing interesting details a Filtering with a kernel size of 5 pixels b 20 pixels c and 40 pixels d Intensity profile plotted shows modulations in the part corresponding to an interesting defect in the part It is obvious that median filter with too large kernel size degrade information gi digisens 70 Figure 3 2 The image a is filtered by a gaussian filter of size 5x5 b
39. by a median filter of size 5x5 c and by an edge preserving with 2 iterations d We can see that the median protects details compared with the gaussian filter and that the edge preserving filter offers performances on top of these two filters while allying the interest of both filters 3 4 2 3 Edge preserving filtering To avoid Gaussian filter limitations blur a filter preserving discontinuities is implemented This filter is recommended if projections have to be filtered keeping a maximum details on object transitions This filter is iterative The only parameter to adjust is the number of iterations in 71 expert mode The more iterations the smoother the images By default the parameter is 2 iterations The figure 3 22 compares the various techniques of filtering 3 4 2 4 Beam hardening compensation The beam hardening coefficient Beam hardening allows the artifacts due to the polychromacity of the source several lengths of wave to be corrected depending on the thickness of the materials passed through by the x rays and the source energy the artifacts appear in the form of an abnormally light grey in certain zones especially around the edges as illustrated in Figure 3 23 The values to use are from 0 no compensation to 10 strong beam hardening For higher values the image may disappear Trony im P Figure 3 23 Beam hardening artifacts before and after compensation 72 3 4 3 Optional Phase contra
40. digisens Gr VOL alu30 Modifi le 15 02 2008 13 39 08 07 2010 11 10 08 07 2010 11 10 8 histogram dat 08 07 2010 11 10 20 Type Dossier de fichiers Document XML DigiECT Volumes Fichier DAT Figure 2 8 Visualization of turbine blades 21 2 4 Volume visualization 2 4 1 Multi display visualization DigiObs offers simultaneously the display of the 3D rendered object and 3 cutting planes Six display modes are available using the menu window or using the button Switch Mode These modes are 3D display 3x1 display 2x2 display oide display Top display Front display The Switch mode button allows the modes to be changed one by one Figure 2 9 shows an example of a 3x1 visualization Se NU 22 WI DigixCT turbine 256 bh30 LE File Window Tools Help Figure 2 9 Management of the main window in 3x1 window mode 2 4 2 Volume slab visualization This mode allows the display of a slab of the loaded 3D volume This mode is only available with volume render activated see preferences menu To enable this mode just click on the button When the mode is active the button becomes yellow Figure 2 10 shows how to change slab thickness using the dedicated parameter Thickness at the bottom of the GUI Slab thickness is in millimeter SE dicisens 23 Figure 2 10 Volume siab display Note When this mode is un adapted with other mode functionalities of D
41. does not exist the field is set to create camera and you will have to create it 1 Hardware definition Camera definition Geometry calibration Geometry definition Create camera iv Pixel size x Figure 3 3 Camera definition tab To manage a camera list the button allows a new sensor to be added the button to remove an imaging sensor Figure 3 4 depicts the camera creation window You have to give to your camera a name its resolution in pixels and the pixel size The third field will be automatically filled in EL la digisens 93 2672 12 457 um amp 12 457 um 24914mm amp 33285 Figure 3 4 Camera creation In some specific applications in order to perform acquisitions of objects larger than the field of view the system is equipped with a shifted detector as depicted figure 3 5 The mandatory condition to perform a good reconstruction with this acquisition setup is to keep an overlapping area between views of at least one third of the size of the detector width When the definition is correct you have to click on Create button to return to the main interface pu X ray I re i Umax 2 u Umax 2 Detector offset l Detector superimposition area u Umax Figure 3 5 Extended view mode Top illustration shows a standard geometry where the orthogonal ray from source to detector hits the detector center for information SOD corresponds to the source to
42. e e Intersection of those three planes Maieisens 35 WI DigixCT turbine 256 bh30 LE File Window Tools Help ume X 5 Y 5 Z 14 ad Repon X Sm Figure 2 26 Volume alignment To fix the orientation of the volume just click the button Lock at the bottom of GUI and move the volume Once the orientation is fixed click again on Lock to validate the orientation To fix the origin of the new reference axis just locate the origin by navigating in the three 2D views on the left of the GUI in figure 2 26 and click on Set button The origin corresponds to the intersection of the three planes When everything is finished all measures described in DigiCUT chapter will be expressed regarding to this reference axis set ps Cg digisens 36 2 6 Data management 2 6 1 Saving and exporting volumes Saving volumes is carried out thru the menu Save Volume of the menu File It can be directly enabled by the icon Fa Saving occurs on the currently displayed volume dimensions resolution display LUT The available formats are a proprietary format vol and two other formats to provide compatibility with third part analysis software Visilog im6 and Volume Graphics vgi formats To save the volumes in another configuration of dimensions the user must use the menu Export volume ofthe menu File This menu allows the displayed volume to be exported thru modification of some of its parameters e Volume dimensi
43. e following commands v Help Displays the help window v Save image Saves the image v Level curve mode Activates the density profile 109 examination mode Y Mesure mode Active le mode de mesure 4 2 4 1 Level curve mode During the analysis of a density profile the line being analyzed is marked by a green line Figure 4 2 A left mouse click allows it to be moved It is possible to switch the analysis with a horizontal line or a vertical line by ticking the vertical box in the contextual menu or not The data can be exported in Excel format with the Export curve data command in the contextual menu Figure 4 5 Example of density profile analysis 4 2 4 2 Measures Activation occurs by right clicking and selecting Measure Mode The activation provokes the appearance of a bar of buttons offering multiple types of measures figure 4 3 Important note this tool proposes informative measure but it does not claim to give a so precise measure as a metrology software would make it The expected precision here will be in relation with the position of the 110 mouse click and to the resolution of the image We shall thus have a precision in the pixel That is the notion of informative measure Note at any moments it is possible to refine the position of the selected points 6 Cut plane Cut Radial Plane Cut Cylinder Cut Cone ok Omm Parameter 0 Omm 1 0 mm Thickness 0 027 mm Reset Stack As
44. e selection min 0 151762 t max 0525745 V Show display range Color presets BE EI 0000000 Add grip Add smooth grip ea Ere MaMigisens Add hard grip Figure 2 17 Color palette m EJ OIS SCIIS P 29 2 4 5 Light sources It is possible to add light sources via the 3 buttons shown in figure 2 18 Figure 2 18 Control buttons for 3D scene illumination modes The effects of these sources on the 3D object are shown in figure 2 19 Adding light sources allows playing with shadows at the same time as with saturation on the part Illumination allows the user to analyze the 3D structure of their part in a more refined way Figure 2 19 Example of applying light source management uH UTSISCTIS Figure 2 20 Example of applying reflection management 2 4 6 Object reflection It is possible to adjust the amount of reflection on the objects It is also possible to have a more metallic rendering of some parts than others This management of reflection is achieved by using 4 buttons shown in figure 2 21 Hence the user can display a mat finish a plastic type finish a metallic finish or a mirror finish Figure 2 21 Control buttons for reflection modes The effects of these sources on the 3D object are shown in figure 2 20 2 5 Managing a volume 2 5 1 Display management Using the mouse It is possible to navigate around the volume object with the help of the mouse see
45. e window size in which triangles are re arranged regarding their neighbors Sharp edge Impact allows preserving regarding its value sharp edges in a surface Each time one clicks on Apply button processing is applied on the current surface Thus you can apply recursive post processing Following figures illustrate smoothing Figure 5 5 Without smoothing C digisens LS m aM e mm C EA entere a ed a thes omo oo m m nr eim an em om emp mm ie m 7 X Figure 5 6 With smoothing region Size 5 00 and Sharp edge impact 0 2 5 2 3 22 Decimation Decimation Decimation reduces the number of 0 triangles triangles defining a mesh One can either JM 0 triangles fixe the desired number of triangles or too fixing a percentage of decimation regarding the initial number of triangles generated Apply launches decimation process on the current mesh Figures 5 7 and 5 8 show decimation result The two post processing can be applied simultaneously Reset processing button allows the restoration of the initial mesh removing all post processing applied Those processing will be lost 134 man UTSISETIS Figure 5 7 Initial surface 1044956 triangles Figure 5 8 Surface with 50 less than initial surface 522478 triangles 135 5 2 3 2 3 Optimization Mesh optimizations are the following e Folding s Angle Threshold allows removing neighborin
46. echnical support ct Graphics extensions report KI Configuration Renderer Intel 45GM OpenGL 1 4 0 Build 4 14 10 4446 Found extensions Found GL EXT bara Found GL VERSIOM 1 4 Found GL ARE fragment program Found GL ARB verter program Found GL EXT testure3D Missing extensions slower display Missing useful GL AHE fragment shader Missing useful GL ARB vertex shader Missing useful GL ARE imaging Extensions required but missing Missing required GL VERSION 1 5 Figure 1 1 Diagnostic window 2 Volume visualization Digi OBS Module DigiOBS module corresponds to the main GUI of Digisens product It allows the visualization of datasets and provides all necessary links between the visualization and product features 12 Maximum resolution depends on the capacity of your workstation In this document we are talking about a standard resolution of 512x512x512 voxels 2 1 Presentation of the default GUI DigiXCT appears in the form of a main window Figure 2 1 made up of a 3D visualization area a menu and a tool bar The tool bar contains short cuis to the main functionalities and parameterization dialogue boxes 2 2 Management of preferences The advanced user has the opportunity to define his environment and a certain number of other parameters via the DigiObs preferences To do So you must open the preferences window which is found in the menu File gt Preferences Figure 2 2 The preferences window di
47. econstruction 5 Preview Adjustment Those 5 steps are mandatory steps to perform a good reconstruction This steps organization provides a clear vertical workflow to the user For each toolbox you will find at least one tab corresponding to one mandatory group box and additional features enabled from expert mode or from license Those tabs provide tools for specific application such as phase contrast tomography or additional features to help reconstructing datasets calibration tool It also gives an access to image processing tools and iterative algorithms For more details on features please contact Digisens team C digisens 51 Section 3 5 will describe a specific use of DigiR3D when the software handles X ray manufacturer system geometry file see system compatibility in this section Figure 3 2 DigiR3D main window 3 3 Hardware definition This toolbox aims to describe user s CT hardware 3 3 1 Mandatory Camera definition In this first tab you have to describe your imaging sensor Several scenarios have been implemented e When DigiR3D is used for the first time the GUI asks you to create a camera see figure 3 3 e When the camera has never been created the software will automatically find the right camera definition and will fill the field in 52 e When several cameras may be compatible with the currently loaded dataset a popup will appear asking to choose the right camera e f the camera
48. ered manner Note The selected spots must be visible in all the images and must never be partially hidden or cut off Important In order for the trajectory of the probe spots to be monitored correctly it is necessary that the initial speed in the image plane of the spots is low i e you start with the spots positioned completely on the left or the right of the image You can reach this position via the cursor below the image It is on this image that the initial position of the probe must be defined Finally the probe can be validated by clicking the short cut Set probe and by returning the distance between the center of the first and the last spot of the probe or by clicking on the button Guess probe if a calibrated probe is used When calibrated probes from Digisens are used and when beads are clicked clicking on Guess Probe will automatically identify the probe and define the distance between the upper and lower spot The accuracy of probe size can be found in all distance measurements of reconstructed volumes an overestimation of 1 160 microns in our example in the probe results in an overestimation of 196 on all size measurements of the reconstituted object 59 POLE EN rrr mma get e a4 YET sanne Figure 3 9 Spot selection Digisens proposes to customers Standard tools a set of two probes One probe is equipped with 0 4mm diameter beads spaced by 1 4mm The second is equipped with 2m
49. figure 2 22 There are two modes of moving one mode for moving the volume and one mode for moving the cutting plane You can switch between these two modes by using the Mouse control mode button of the tool bar or by depressing the spacebar on the keyboard The icon 3D volume R becomes cutting plane mode ES The cutting plane is moved by manipulating the mouse in the same manner as for moving the volume An example of a cutting plane is shown in figure 2 23 The left button of the mouse allows the cutting plane to be turned The right button allows translation in the plane and the roller wheel allows perpendicular translation to the plane Keyboard key strokes can also be used and are shown in figure 2 24 left double click left double click reset view reset cutting plane left double click right set rotation center zoom in Azoom out volum translation cutting plane translation volume rotat cutting plane tation SPACE BAR TOGGLES MOUSE MODE volume control cutting plane control Figure 2 22 Use of the mouse for the manipulation of 3D volumes lg digisens 32 Figure 2 23 Use of cutting plane mee Keyboard shortcuts F display frequency G display grid ALT ENTER toggle fullscreen D screenshot C show hide outlines SPACE toggle volume cutting plane modes b Figure 2 24 Details on active key strokes 33 2 5 2 Volume alignment Displayed volumes are expres
50. g triangles for which the angle between their normal vectors is equal or higher than the defined parameter This option allows for instance to remove folds in meshes e Small Volume Threshold allows removing meshes delimiting a volume lesser or equal to the defined parameter e Remeshing This option allows you to perform a remeshing in order to get triangles shape close to equilateral triangles This computation may be very long regarding the number of triangles to process A rough estimation of the processing time is indicated 5 2 3 3 Display properties Display properties Display properties allow the user to choose Color C Internal color the external color of the mesh color as paciy 4 wel as its internal color Internal Color It also allows defining transparency of the surface Opacity to display it using a wireframe rendering to apply clipping plane onto it Enable Clip Wireframe Enable clip Figure 5 9 Note clipping plane is independant for each extracted surface x MN 136 mg B a aet en PSS HSE Sete N yn RS an war TL RL LAS ANA LS AMEN pea TET No c meg NOTES NS SEBASTIAN SA NY NG 4 ij SA A RY KEN UN LIRE ETES Se NA AR WA ANA d EK UND RR TAS VUA III KIROT DALER E AN 4 IN Al Dar IS pe L gt ASLAN ASI BEDS S R 217 47 ASD PA Tie d v r kt V R oca OF TE VEM zd VA k EL Pa M f MU Wa TIAN a 7 LIE
51. he axis of rotation Outside data corresponds only to the intersection of some beam rays common to some projections They are thus carriers of information with weaker signal to noise ratio The information is thus degraded by a stronger noise and the levels of density can be biased The figure 3 36 illustrates the effect of the cylindrical constraint 83 Perspective Perspective Figure 3 36 Effects of the cylindrical constraint on reconstruction Top without cylindrical constraint bottom with cylindrical constraint 84 3 6 6 Reconstruction grid orientation In expert mode it is possible to rotate the volume according to an orientation chosen during the preview step see figure 3 37 As seen in the chapter 2 5 several axis references are used in the software The modification of the quaternion acts directly on the volume reference To modify the orientation of the object just fix the axis of rotation and the value of the angle of rotation J Manual grid rotation Angle degree 04 Axis X mm 0 Y mm 0 4 Z mm 0 5 Figure 3 37 Rotation grid definition During the modification of the grid rotation parameters an overlapping box is added in which it can be seen the volume preview This additional box illustrates the rotation that will be applied to the reconstruction grid with regard to the original reference 3 Heconstruction mandatory Digisens software allows you to select used reconstruction algorithm
52. header vol The min point the max point and the center of the volume can potentially be modified with the actual dimensions of the stack of images You can also specify the resolution of a volume 256x256x256 by default and the size of a voxel in mm The Quick scale buttons allows volume parameters to be set more quickly either in full resolution or in a lower quality resolution The import is achieved by clicking on Proceed Once the calculation is finished a volume file vol is created in the pre selected directory It is possible from this new volume file to use all of the other 3D treatment plug ins and to export 2D slices in other directions Note the imported slice name has to be numbered with the same number of digits For example if the last slice number is slice 532 tif the first slices have to be numbered slice 0000 tif slice OO0Ll TtIET 143 7 Automation of reconstruction tasks Batch Mode As with all the plug ins in the DigiXCT software suite the batch mode is protected by a software key 7 1 General presentation DigiXCT offers the user the possibility of carrying out a group of tasks automatically and without supervision Batch mode consists of writing a small command script which requests the execution of basic tasks such as reconstruction display or even other functions 7 2 Batch mode functioning To access batch mode you only need to launch the command of the Window gt Script menu repre
53. ich allows Calibration of tomography systems Reconstruction in different geometry configurations Visualization of volumes and meshes Manipulation of a 3D cutting planes Analysis of density levels based on histogram Basic measurements The software is made up of 2 main modules and processing plug ins The two main modules are DigiOBS Visualization and 3D analysis software DigiXCT DigiR3D Reconstruction software Its plug ins are CUT Analysis of cutting plane amp measurements SHF Extraction and mesh visualization SIM Import of slices Batch Mode Module for producing macros by command line The modules and plug ins are brought together in the same interface based on the DigiObs module 1 2 Minimum configuration requirements Minimum recommended configuration for a desktop computer e Double slot PCle motherboard e Intel quad core processor e 12Gbits RAM minimum two times the total amount of GPU RAM e Windows XP Windows 7 64bits 10 e 10000 rpm hard disks mounted as RAIDO drive e NVIDIA Fermi board dedicated to computation Tesla board e NVIDIA board for visualization Minimum recommended configuration for a laptop computer e Intel Dual Core Processor e 4Go RAM e Windows XP Windows 7 64bits e High speed hard disk e Quadro 2000M graphic board e NOIGP ENABLED e OPTIMUS TECHNOLOGY DISABLED For efficient and optimized configuration please contact Digisens technical team at support digisens3d com
54. igiXCT the activation button is grayed 8 2 4 3 Display parameters To visualize a volume it is necessary to choose the range of density values that will be displayed It is possible that the visualization of a volume is subject to interference such as noise or un adapted display LUT Figure 2 11 In such a case you must click on the volume Appearance icon on the tool bar in MISSIS 24 order to open the parameterization window The diagram shown is the histogram of the object s density lt indicates in abscissa the density values describing the volume and as coordinates the number of voxels parts making up the volume having these densities To modify the density range you have to move the minimum and maximum limits around the zone symbolizing the object either by moving the hard grip in the diagram or by modifying the min and max values shown below the diagram see figure 2 12 It involves the second largest zone the largest symbolizing the spike in the data curve In order to refine this zone you can zoom in and out between the min and max ranges by choosing the option Show display range Figure 2 13 The volume appearance is updated automatically i 1 2 7 Show display range a ex Ne x y Color presets Y BE B0000000 Add gro Figure 2 11 Wrong density range uH UTSISCTIS Ml DigiXCT turbine 256 bh30 LE ts File Window Tools Help Ws UE amp amp 0 Fe Aa i HEG
55. igure 2 30 You only need to specify the name of the AVI file or the base name of the images in the corresponding field You then have to click on the Record button to start the recording and then on the stop button to stop it A new window offers a choice of compression codec available in your computer registry directory The film thus produced is the output of the 3D zone and of all the events that influenced it during the recording 40 This function can also be activated via the button W m Movie record me AVI name save images Configure default AVI format m d Figure 2 30 Film export dialogue box Note It is possible to remove cube borders in the 3D visualization area symbolizing the inclusive volume or to display it with pressing the c key It is possible to record a high resolution animation The principle is quite simple the function allows you to compute the entire trajectory of a camera moving around or inside the object from discrete camera locations To do so some keyboard keys have to be used e Alt F1 save the actual camera location a message appears bottom left saying Position 1 saved e Alt F1 twice erase the entire path saved e F1 play stop the animation 2 6 4 Screen dumps The command Window gt Export Image allows a screen dump of the current display to be carried out Once the mode has been activated you only have to choose the directory the name of the file to be saved
56. in order to reconstruct an optimal volume in relation to the data supplied or they can even be adjusted manually in order to only reconstruct a chosen zone Figure 3 31 3D reconstructed volume definition The button AutosizeGrid displays a menu with 3 options described below Minimum reconstruction algorithm will take into account only regions in the volume visible in all projections Minimum Half illuminated reconstruction Maximum possible reconstruction 79 Half illuminated reconstruction algorithm will take into account regions in the volume that are visible in at least half of the dataset Maximum possible reconstruction the entire volume will be reconstructed l I I I I I I I I I I L Jn9o32919Qq Source I1NIPI IG Jn9o32919q Figure 3 31 Autosize volume definition From the top Maximum possible reconstruction Half Illuminated Reconstruction Minimum Note when export mode is invalidated Autosize grid button fixes dimensions to Half Illuminated Reconstruction Note intuitively we could expect that default dimensions proposed by the software have in width and height the resolution of the detector The figure 3 33 illustrates the principle of estimation of the dimensions In a conical context the reconstruction will be made in the cylinder registered in the cone of the beam defined by the dimensions of the detector The reconstruction grid contai
57. ing is slower but finer Sharpness filter Iterations 3 Radius 3 Speed 05 Figure 3 46 Sharpness filter parameters Figure 3 47 Effects of sharpness filter reconstructed slice of a granite sample 3 8 Preview and Adjustment 3 8 1 Preview Settings Preview reconstruction allows the parameters entered in previous stages to be quickly checked and possibly to be adjusted iteratively ie 98 adjusting the trajectory definition or even the dimensions of the volume to be reconstructed DigiR3D proposes two preview modes 1 reconstruction of the whole 3D object in low resolution 2 reconstruction of a single high resolution slice which is arbitrary placed thanks to the cutting plane The two modes pre calculate a sub collection of the low resolution filtered projections in low resolution to accelerate the successive previews it is this sub collection that we refer to as cache It is possible in the Images for preview of the dialogue box to modify the number of projections to be used as well as their resolution by default adapted to the size of the low resolution volume The limit which prevents the use of all of the projections in full resolution is the memory used by this cache which is displayed in memory usea for information Low resolution volume preview IHA Low resolution preview is accessible whichever tab is displayed at the bottom of the dialogue box To launch the reconstruction one has just to c
58. ion geometry tab 57 Steps for calibration e 1 Image loading see figure 3 7 To choose the series of images to be worked on you only have to click on the first short cut of the toolbar Choose input images All images of a similar name base name without image number will then be selected 1 Hardware definition Camera definition Geometry calibration Geometry definition Probe images C DATA XCT Delpx 20110706 2011_07_06_15 19 19_mire demo ima0001 tf Autocontrast White 255 Gamma 128 Black 0 Probe Guess probe Set probe Initial search radius pix 05 Constraints Region of interest 7 Fixed focal length 0 mm Prem geet Figure 3 7 Image loading for geometry calibration e 2 Probe definition The Probe tool see figure 3 8 allows defining the size of the probe the relative position and number of spots Probe Guess probe Set probe Initial search radius pix 0 8 spots 78 768mm CAL 001 0 n Al NS amhe LAL UU I UUUUT probe v Figure 3 8 Probe definition 58 First it is necessary to indicate the dimensions of the spots in the initial search size field A left click on the image of a spot launches the automatic search of its center and its radius A click on an already selected spot deselects it And a double click deselects all the spots Note The spots must be selected in order preferable from the highest to the lowest on the screen and never in a disord
59. ir resolution on the quality of previews The slice on the left has been calculated based on 45 images of 128x160 pixels reusing the same cache as the precedent volume preview The slice on the right has been calculated based on 180 images one out of two of 768x960 pixels full resolution 3 8 2 Adjust Although geometric parameters or beam hardening coefficient can be finely adjusted thru the previews in practice it can turn out to be tedious to wait for the end of the calculations during multiple iterations on a given value It can also be impossible to get a sufficiently accurate preview particularly because of memory limitation which prevents the use of all the full resolution radiographs available in the preview cache kasa MAG eye ee 102 Figure 3 50 Parameters list Preview settings Parameter adjustment Metal artefact reduction Parameter s mia Theta radius 1 deg White 14 Black og Choose best image position F Autocontrast Note post processing fiters are not supported during adjustment and will be silently ignored Figure 3 51 Adjustment tab GUI For these two reasons a semi automatic tool is proposed see figure 3 51 The user chooses on which variable he wants to make the adjustments Then the parameter increment has to be set up step By clicking on Recompute button the software computes the slices defined in Preview Settings tab with the following parameter values e Init
60. le to remove it from display by un ticking the show Preview box Preview by high resolution slice A Preview by slice is only available if the ART reconstruction is not selected in Preferences Preview by slice is accessible regardless of the tab displayed at the bottom of the dialogue box To launch the reconstruction one has just to click on the button iJ It allows the reconstruction of a slice arbitrarily placed in a greater resolution 1024x1024 by default The position and orientation of this slice can be set in the dialogue box by specifying a Plane center point and a normal to this plane Plane normal It can also be set by the normal cutting plane which can be manipulated in 3D display for example on low resolution volume preview Note The successive slice previews open a new display window each time to allow comparisons between the different slices versions Note It is possible to interactively edit the title of the window to put a more appropriate comment in it on the parameters being modified for example By default the title re uses the position of the calculated slice In following figure we can see two 512x512 slices calculated with two sub collections of different radiographs SALSA LLI 101 Figure 3 49 Preview of a 512x512 pixel slice with 30 128x160 pixel radiographs on the left and 180 768x960 radiographs on the right This illustrates the importance of the number of radiographs used and the
61. lick on the button IA It reconstructs the corresponding volume in a few seconds By default three resolutions are offered 128 256 and 500 Thus we have a visualization which is close to that which will be reconstructed see figure 3 48 Preview settings Parameter adjustment Metal artefact reduction Input projections Images for preview KA Images resolution 50 0 384x480 iv memory used 63 3 MB Low resolution 3D preview Voxels in the preview grid 256x256x256 v Preview grid memory 64 0 MB Slice preview Slice resolution 1024x1024 iv Plane center 0 3 4 0 E Plane normal 05 0 14 Preview slice memory 4 0 MB Figure 3 48 Preview parameters It is then possible to quickly check the coherence of the entered parameters and to adjust the volume reconstructed on a possible area of interest Note lt is possible to manually enter a reconstruction resolution such as for example 32x128x64 However as the reconstructed volume is only destined for 3D display we recommend that you follow the restrictions imposed by the graphics cards used the resolutions should be a power of two and for most graphics cards the 3D display resolution is limited to 512 Not following these restrictions does not involve an error but causes re sampling useless for the display with respect for these restrictions 100 Note volume preview remains displayed even after the closure of the dialogue box It is possib
62. lug in As with all the plug ins in the DigiXCT software suite DigiSRF is protected by a software key 5 1 General presentation of DigiSRF SHF is the DigiXCT software suite s surface extraction plug in SRF calculates a surface corresponding to a given level of density based on a volume coming from a tomographical reconstruction It is based on the Marching Cubes algorithm This plug in appears in the form of a dialogue box which is accessible via the Tools gt Surfaces menu or via the button e 5 2 DigiSRF functionalities 5 2 1 Surface manager SHF s main dialogue box Figure 5 1 allows surfaces to be created and removed the volume visualization to be masked and the global parameters concerning optimization of calculated surfaces to be adjusted The New button allows a new surface to be created It opens a new tab see Figure 5 1 By default the new surfaces are empty and nothing is displayed on the 3D display Refer to 5 2 2 Surface Operations for the effective creation in 3D display and adjustments of a surface The Delete button destroys the selected surface En vocanc W digisens 127 Figure 5 1 Main dialogue box of SRF plug in 5 2 2 Surface operations Each created surface corresponds to a tab figure 5 1 By default surfaces are named surface i Necessary parameters to define a surface are the choice of an isovalue defining iso density surfaces iso value post processings applied onto the
63. m diameter beads spaced by 5mm As an option calibrated probes with metrology inspection report see figure 3 10 The probe and the report are unique Those probes have two advantages the tomography system is linked to metrology national references they provide a simple tool for system calibration follow up The design is unique Thus calibration tool is able to automatically detect spots and provide the user with the right probe size using the button Guess probe Using such tool provides a calibration repeatability lesser than 5um whatever system focal and user settings i digisens 60 r f document QUALG A Calibration certificate AM2008 809 Figure 3 10 Calibrated probes e 3 Optimizations optional The Region of interest section see red and green borders figure 3 12 allows the image to be cropped which is rarely useful other than when some spots get too close to an image border leading to a tracking failure The uMin uMax values therefore represent the horizontal interval used and vMin vMax the vertical interval It is also possible to fix focal distance value as a prior knowledge to improve geometry estimation accuracy e 4 Geometry estimation Once the probe has been specified the Compute button allows the calibration calculation to be launched This calculation can fail red marking on one or several spots for the following reasons 61 v The initial speed of the spots is too high the
64. n the 3D view window This superimposed object displays the plane used to perform slice extraction There are two ways to change the plane orientation and the location either axis and base coordinates are entered in the corresponding fields or the user uses the mouse To use the mouse click on R or press the space bar and use the mouse as follows v Right click Plane axis shift v Left click Plane axis orientation change To extract a radial cut first fix the following values e Frame width width Frame height Height Revolution axis Axis Frame axis base Base Start angle for radial cut Start Angle ll DANE PET iw 125 When parameters are fixed click on Apply button to get the corresponding slice DigiXCT alu300_v2LE Perspective Cutplane Cut Radial Plane Cut Cylinder Cut Cone FECE Figure 4 13 Mode coupe radial Cut Radial Plane Buttons Back and Next makes the cutting plane moves around its axis with an angular step equal to Thickness Views are automatically recomputed Button Stack As grabs a set of images with a rotation range of 180 degrees Each image corresponds to a cone base radius increased by a radius step equal to Thickness The user can set the origin of the extraction first image by fixing the Start Angle value from 0 to 360 degrees The Reset button relocates the cylinder axis and base to initial values 126 5 Visualization and extraction of surfaces DigiSRF p
65. n the computer It corresponds to the remaining memory once all the system processes have been launched CPU Reconstruction 15 The GPU Reconstruction check box enables the use of GPU s for Manual selection of GPU s reconstru ction 4 GPU Reconstruction This menu also allows the user to explicitly choose GPU board s used for reconstruction using SnapCT among a list of all available Cuda compatible devices 2 2 2 Interface tab DigiXCT proposes a tool for informative measurements see DigiCUT Measure unis plugin To be adapted to the system of Mai n units of one s country it is possible to convert measures in mm international 9 j system of units or in imperial system of unit ie inch mils Important note This change in units is not applied to tomography geometry definition that remains in the international system of units The user can by using this interface Figure 2 4 change the background colors of DigiObs He can either use a plain color by putting the same color in Background color up and Background color down or create gradients nahati Textured option allows the use of an NY O image as wallpaper for DigiOBS Compatible formats are jpg bmp or png This image replaces colors defined in Color Up and Color Down The color Selected color corresponds to the color of the elements selected in the 3D grid and Outline color to the color of the physical limits of the visualized 3
66. ns this cylinder The figure shows that in a general conical case the cube is narrower here than what it would be if the beam divergence was weaker because the inscribed circle has a diameter smaller than in a case where the beam divergence is weak 80 Figure 3 33 Determination of reconstruction grid dimensions 3 6 3 Region of interest You can reconstruct a region of interest in your volume see figure 3 34 To do this you can either return the dimensions of the area of interest using the DigiR3D interface or by using the multi display visualization described in chapter 2 4 and defining the area of interest by using the mouse In this latter case the Define ROI for reconstruction box must be ticked The dimensions of the region can simply be changed by left clicking on the boundaries and whilst keeping the button pressed down gently dragging to change the boundaries The 3D visualization window which defines the volume is thus updated To refine the definition of the area of interest it is possible to zoom in by pressing the keyboard s ctrl button and by using the mouse s roller wheel Returning to the original zoom is achieved by pressing on the keyboard s ctr1 button and by left double clicking on the mouse 3 Output Output format 2 Volume v2 A Output fil 1 1 wd FA LE A A AA ENL A nes ON AR NEM ES HA Dimension Autosize grid X Y Fi Center 4 3 2139 mm 4 716137 mm 13 329869 mm Size 12 098 mm
67. of a region of interest of a human jaw teeth close up Bottom left image is a reconstruction without SART algorithm bottom middle image with SIRT algorithm bottom right image with NITRO The two first algorithms have the well known truncation artifact e The regularization depicts a minimization scheme SNN filter or ICM minimization with a total criterion variation during the 91 reconstruction allowing the introduction of prior knowledge on the volume ie piece wise constant objects The effect of the regularization is balanced by the factor Lambaa 0 11 In comparison with analytical algorithms iterative algorithms e Are more robust in relation to noise e With fewer limited view angle artifacts e Are slower necessity to iterate e he whole volume needs to be stored in the memory In practice algorithm SIRT will privilege a good signal to noise ratio SNR regardless to resolution while algorithm SART will make the opposite Algorithm OSSART will be between the both The ART mode is only available with GPU reconstruction The multi GPU reconstruction is possible if the volume can be stored on all GPUs available in the reconstruction system The particular case of the parallel geometry with vertical axis allows the use of multi GPU without restrictions Important note because of iterative specific implementation the largest reconstructed volume corresponds to half of the total GPU memory one third if 3D post proce
68. ons Figure 2 27 by ticking Resize it is possible to choose the final dimension of the output volume Wl Export Slices and Volume As Figure 2 27 Modification of data dimensions EL la dicisens 37 e Output volume format Figure 2 28 The volume can be written in the current format of the volume v2 either in a format compatible with Volume Graphics VGMax or VGStudio software or the Visilog software Wl Export Slices and Volume As V2 Volume C vo Figure 2 28 Export formats 2 6 2 Exporting slices Files containing the description of a volume have the extension vol The DigiObs module allows the visualization of such a volume in 3D with reduced resolution It is possible to extract a series of slices in different file formats bmp tif raw jpeg Slices can be exported according to the X Y Z orientations For custom orientations see chapter on DigiCUT plug in MG tapon Sees ond ohne Aa are Figure 2 29 Slice extraction dialogue box You must choose File gt Export Volume in the menu Then tick Export Slices Figure 2 29 To carry out a slice extraction in the Base name you must specify a destination directory and a root for the name of the extracted images If you choose c object slices slices_ in the c object slices directory you will get a series of images called slices 0001 bmp slices 002 bmp It is also necessary to specify the direction of the slices X Y Z and whether
69. r than 1h According to scan duration this drift is assumed to be linear In this case it is easy to compensate its effect using an extra image acquired at 360 and using a linear compensation model For very long scans drift is more complex and need more than one image To use this drift compensation load the extra image in the dedicated field and click on compute button to estimate de drift magnitude Then all projections will be shifted to compensate from this drift If one wants to cancel the compensation just click on clear button X ray spot drift C DATA XCT extra mage tif al Compute Clear Select an extra projection to adjust tube drift Figure 3 28 X ray spot drift correction parameters 76 Note the software is compatible with Skyscan drift compensation To use all those extra images just load the first iif file from Skyscan dataset 3 5 X ray CT system compatibility DigiR3D is compatible with the following CT system manufacturers GE Measurement amp control solutions Phoenix GE Healthcare for small animals imaging Skyscan Nikon Metrology XTek RX Solution Cyx Yxlon NSI For a detailed list of compatibility please contact Digisens commercial team If the CT is handled by DigiXCT when the user clicks on load the file browser opens and the user will be able to choose the geometry file associated to the CT system as depicted in figure 3 29 If the system is not handled by the software
70. rform slice extraction i digisens 120 There are two ways to change the cylinder orientation and the base either axis and base coordinates are entered in the corresponding fields or the user uses the mouse To use the mouse click on R or press the space bar and use the mouse as follows v Right click Cylinder axis shift v Left click Cylinder axis orientation change REDUCES EE Cut plane Cut Radial Plane Cut Cyinder Cut Cone amp Parameter Figure 4 10 Cut Cylinder module To extract a slice first fix the following values e Base radius R Cylinder height Height Hevolution axis Axis Cylinder base origin Base Maximum radius for stack extraction Rmax When parameters are fixed click on Apply button to get the corresponding slice 121 Note regarding object dimensions the computation may take several seconds Buttons lt lt Back and Next gt gt makes the cutting plane moves along the normal axes direction with a moving step equal to Thickness Views are automatically recomputed Button Stack As grabs a set of images Each image corresponds to a cylinder base radius increased by a radius step equal to Thickness The base radius range stats at R and finishes at Rmax Button Slice As grabs only the current displayed slice Note the value Rmax is used to fix slice image dimensions The user can set the origin of the image first pixel by fixing the Start Angle value This value acts like an off
71. rgent beam horizontally and parallel beam vertically 3 4 Input This second toolbox aims to describe and to process projections used to achieve a correct reconstruction 3 4 1 Mandatory Input definition The Input files field see figure 3 18 indicates the directory where the projections are located that can also be selected using the explorer by clicking on the button h The image formats supported are TIF DICOM VFF ATN RAW JPG BMP m digisens 65 digiR3D Registration and reconstruction plugin Figure 3 17 Input tab presentation Image files Attenuation images Use flatfield Set white level n Figure 3 18 Input files According to acquired image type several scenarios have been implemented If images are intensity based images you have to use a flat field image also known as bright image beam profile image To 66 use it and to inverse the Beer Lambert law just select it using the explorer by clicking on the button b Automatically the software will compute an attenuation image by dividing the flat field image by the acquired projection and by computing the logarithm of the resulting image e If the flat field image does not exist you can generate a constant image with a level corresponding to the mean value of the bright image This level can be measured on a small region of interest where the object is not present or you can use the
72. s regarding to the geometry of the acquisition system or scan set up Hence you can choose among standard reconstruction methods such as FBP FDK and iterative approached leading to improved image quality in strongly degraded context low SNR low contrast limited view angles few projections To select any reconstruction algorithm you just have to click on the arrow of the combo box to display the list of available methods see figure 3 38 85 Method Volume processing Scaling Method Analytical method parameters Windowing HAMMING vi Double padding Figure 3 38 Reconstruction methoos list 3 7 1 Filtered back projection Filtered back projection is the most known approach in tomography reconstruction Its ease of use and speed make it a reference in the field of application This approach is a member of analytical reconstruction methods Regarding the beam type See chapter 3 3 3 the list shown figure 3 16 is updated form FDK cone beam to FBP parallel beam FDK or FBP need the filtering of projections by the mean of a ramp filter defined in the frequency domain The main drawback of the ramp filter is that it amplifies noise as it restores object resolution In order to reduce noise degradation in reconstructed volume an apodisation filter is combined to ramp filter lis aim is to reduce or to eliminate high frequencies in the image noise Unfortunately high frequencies are necessary to restore the object
73. sed in a direct reference axis which center coincides with the center of rotation of the system This reference axis called world reference figure 2 25 can be activated with the keyboard shortcut a It is used to determine the center and the orientation of a section extracted from the volume Digi OBS uses two other reference frames e Areference frame in relation with the region of interest described in the chapter about DigiR3D and named volume reference figure 2 25 e reference frame for measures allowing to express them regarding to an alignment that can be in relation with CAD model measure reference in figure 2 25 It is thus possible to align the volume by moving and or rotating the measure reference This feature is very useful in measure mode because it will allow expressing the location of an entity with regard to an absolute reference which can be a CAD reference for example To enable this mode click on the button A When the mode is active the button becomes yellow em and fields appear at the bottom of the GUI figure 2 26 Note This mode may be un adapted with other mode functionalities of DigiXCT In this case the button is grayed a C digisens 34 o PE Reference Reference Figure 2 25 Reference axis definition in DigiXCT During alignment process the former orientation of the volume remains displayed with dashed lines The new reference is defined by e Cutting planes axis red green and blu
74. sented by the icon The window shown in figure 7 1 appears The list of basic functions can be obtained by hitting help in the script window then clicking on the Go button Figure 7 1 Script window Figure 7 2 shows a script example which allows e Loading a calibration file Em digisens Requesting a reconstruction with interpolation of the voxels Set the final resolution at 256x256x256 Launch the reconstruction via DigiFDK Display the volume in DigiObs 144 Execution occurs by clicking on the Go button Figure 7 2 Example of script program
75. set from 0 to 360 degrees along the perimeter of the base The Reset button relocates the cylinder axis and base to initial values Note in this mode the image size is automatically computed and is not editable lisa KA D a E E aJ 122 4 3 3 Cut Cone This mode is an option of DigiCUT Please contact Digisens for further information When Cut Cone tab is selected Figure 4 11 a transparent cone and its axis of revolution are displayed in the 3D view window This superimposed object displays the curvilinear surface used to perform slice extraction Wl DigixCT turbine 256 bh30 LE EEECE File Window Tools Help E Perspective FR fs Seas os AAAH Cutplane Cut Radial Plane Cut Cyinder Cut Cone arameter J 06 Base P UOI4 Fe N4mmi Rtop Omm Start Angle O deg Jeu 05 133mm t Rdown 18 288 mm 2 Height 39 939 mm 114mm 2 Rspan 0 037mm gt Thickness 0 148 mm Figure 4 11 Cut Cone module There are two ways to change the cone orientation and the base either axis and base coordinates are entered in the corresponding fields or the user uses the mouse To use the mouse click on Q or press the space bar and use the mouse as follows v Right click Cone axis shift v Left click Cone axis orientation change 123 To extract a slice first fix the following values e Bottom base radius Rdown Top base radius Rtop Cone height Height Revolution axis Axis Cone
76. splays 3 tabs Computation parameters useful for tomography reconstruction e Interface user interface color parameters management of windows and multi display configuration e Graphics display parameters mi dicisens 13 L M A File Window Tools Help Input Output Visualization Plugins buttons management management Visualization area ghligsen Figure 2 1 DigiXCT main window Window Tools Help Open Volume 4 Save Volume P Export Volume and Slices Dreferencec Frererences Volume Info Exit Figure 2 2 Opening of the preferences window 14 2 2 1 Computation tab This tab has two fields used for reconstruction These parameters are straightforwardly used by DigiR3D plug in CPU Reconstruction and GPU Reconstruction Figure 2 3 m DigiOBS global preferences Computation Interface Graphics Memory limit 6 4GB CPU Reconstruction J GPU Reconstruction Manual selection of GP U s Figure 2 3 Preferences window computation tab The number of threads for computation corresponds to the number of processes that can be launched in parallel onto the CPU It depends on the workstation s architecture and more specifically on the total amount of cores in processors The preferred memory size for the calculation of a sub grid corresponds to the maximum memory that can be allocated for the calculation This value depends on the amount of RAM memory available o
77. ssing is used Three reconstruction schemes are available in the list Method see figure 3 38 SART SIRT OSSART If the OSSART is chosen the user defines the value of the parameter Projections Subset It is possible to modify the value of the parameters Iteration and Convergence Speed 92 The number of iterations and the value of Convergence Speed parameter can provide optimal convergence whatever the method chosen see figure 3 39 aa Sil it livergence X Slower convergence 8 Y I Optimal v Convergence I 1 mer teens Figure 3 39 Iterative algorithm convergence To improve image quality the user can use post processing described in chapter 3 7 3 but also specific features for iterative algorithms These features are Regularization the expert user can use regularization schemes into the process of reconstruction in order to increase the signal to noise ratio and to emphasize details of the reconstructed objects He will have the choice between Symmetric Nearest Neighbor SNN and Total Variation TV figure 3 40 93 Volume nriors z Regularization Method SNN Strenath ID Constraint to positive values Figure 3 3 Regularization parameters The Strength parameter which is used as a weighting factor corresponds to the balance between prior knowledge and measurement Ideally if its value is 100 as much of weight to the regularization that to
78. st pre processing Phase contrast pre processing allows the reconstruction of a set of mixed projections This kind of projection contains both absorption and phase contrast A mixed projection can be easily recognized when object components edges are brighter than regions surrounding Without any processing this phase component introduces severe artifacts in reconstruction when restoring object edges See figure 3 26 The implemented approach is based on Bronnikov based correction This module is under license Image correction has to be performed in two steps e First step apply Bronnikov filter on image to split pure absorption signal from phase signal Choose a value that leads to a low frequency representation of the original image See figure 3 24 3 25 e Second step fix the BAC correction such as you are close to contrast inversion and with a value lower than MBA correction parameter Compared to BAC image this one does not show any phase effect and high frequency information is included See figure 3 24 3 25 Input definition Pre processing Phase contrast Image correction MBA Correction 9000 White 0 59 BAC Correction 8200 Black 0 88 4 Auto contrast Figure 3 24 Phase contrast filtering parameters UISISETS 73 Figure 3 25 Effects of MBA and BAC corrections on the projection of a carbon fiber plate Figure 3 26 shows the corresponding reconstruction with and without phase compensation 3 4 4
79. sualization and for further analysis In order to achieve good results it is important to apply the following rules v Geometric settings of the CT system must not be changed between stages 1 and 2 i e the position of the tray the source and the detector must be maintained Because of the mechanical tolerances of the robotics the robotic arm never returns to the requested position even if the accuracy of the positioning sensor is close to one micron Additionally if the object to be reconstructed is heavy it is appropriate to put in a tare of approximately the same weight to compensate for the effects of robotic buckling The buckling implies variations in the position of the rotation axis of several microns otages 1 and 2 can be executed in a different order A whole series of objects can be scanned with the same calibration but it is nonetheless prudent in this case to do at least calibrations at the start and the end of the acquisition series so as to ensure that the machine has not shifted or been accidentally adjusted At least 5 spots must be chosen 10 are best They must be at least one diameter s distance away from the edge of the image and remain so during the whole revolution 3 3 2 2 Calibration GUI To reach the calibration interface click on the tab Geometry calibration See figure 3 6 1 Hardware definition Camera definition Geometry calibration Geometry definiti Figure 3 6 Calibrat
80. the user can fill all necessary fields based on the procedure described from chapter 3 3 to chapter 3 4 mi dicisens 17 je Ordinateur OS C DATA HCT Crane Reconstruction Organiser v Nouveau dossier z 2 Nom Modifi le Favoris RE Bureau crane 256V2 vol 21 11 2011 14 19 Dossier de fichiers T Emplacements r B T l chargement C3 Biblioth ques 3 Documents ii Images 3 Musique Bi vid o a Groupe r sidentiel 18 Ordinateur amp os C Ga R seau Nom du fichier L GE CT 120 scan xml Figure 3 29 Compatible CT system list 3 6 Mandatory Output definition 3 6 1 Volume format and name Two formats are handled by DigiR3D e Image stack format format vO with extensions tif raw dicom jpg bmp e Digisens 3D file format volume v2 Figure 3 30 Projections registration lt 78 To define the name and the location of saving of reconstruction file you have to click on the button I to open a file browser and write necessary information 3 6 2 Volume dimensions to be reconstructed The Dimension field allows the name of the file containing the reconstruction data and the directory in which they will be stored to be specified The volume dimensions in mm are defined by the center of the 3D grid to reconstruct Center and the size of the grid Size shown in red see figure 3 31 These two values can be automatically positioned auto box
81. tween the two segments defined by first second points selected and second third points selected ical ta Cut plane Cut Radial Plane Cut Cylinder Cut Cone ok Omm l Parameter 05 Omm TE Omm Thickness 0 027 mm 3 Reset Stack As Sice As lt lt Prev Apply _ Next gt gt Resolution Contrast Zoom Image size white 3 436 x r 8 gamma 112 t x SE Lad black 0 708 SJ LJ Pixel size 0 0266833 mm Volume contrast 1 1 V JL Figure 4 6 Angle measure 4 2 4 2 5 Circle measure To measure a circle two methods are available e First method o Select circle measure P o Verify that the assisting mode is deactivated 115 o Left click to select the circle center o Left click to determine circle radius Displayed information is e Circle diameter e Location of the center in the reference e Perimeter value e Disk area delimited by the circle e Second method o Select circle measure N o Select the assisting mode L it is possible to deactivate it o Left click on 10 points defining roughly the circle you want to define o When the tenth point is selected the circle is adjusted to fit the object border Displayed information is e Circle diameter e Location of the center in the reference e Perimeter value e Disk area delimited by the circle 116 Cut plane Cut Radial Plane Cut Cylinder Cut Cone ok Omm Parameter 18 Omm DE Omm Thickness 0 027 mm
82. y 02 Compute J Post processing Smoothing Region size 54 Sharp edge mpact 02 L Apply Decimation 1044956 gt tnanges 100 triangles 2 m igisens Figure 5 11 Mesh extraction using Dual Contouring approach C digisens 138 5 2 4 Simultaneous visualization of multiple surfaces It is possible to compute several meshes on a same volume by changing iso value For the example presented in this chapter we can compute a second surface with the iso value 0 390 figure 5 12 Post processing Smoothing Region size Sharp edge impact Figure 5 12 Second surface extraction It is possible to visualize surfaces simultaneously using transparency parameter or and using clipping plane Figure 5 13 SE digisens 139 5 3 Saving reloading Each tab linked to a surface is in association with two buttons Load and Save allowing to save and reload a surface Possible file formats are available in read write mode are 1 Internal format surf 2 OBJ Wavefront with extension obj 3 SIL ascii with extension st Figure 5 13 Visualization of two surfaces changing transparency 140 6 Import of tomography slices DigiSIM plug in As with all the plug ins in the DigiXCT software suite DigiSIM is protected by a software key 6 1 General presentation of DigiSIM This plug in allows existing tomography slices to be imported parameters for sizes and corresponding resolutions to be set and the slices for a

DIGIXCT SOFTWARE USER MANUAL

Contents

Download Pdf Manuals

Related Search

Related Contents