Home

ECG2HRV: Electrocardiagram to Heart Rate Variability - E

1. MMConMenuPrincipal 1 tucontuarzoo iffConHRVAnalyss ItfVisRRDetection Figure 5 10 Component Diagram of RRDetection control 5 6 6 HRV Analysis Brief Description This component is responsible for performing and displaying the results of HRV analysis through a visual interface It contains the implementation of the signal processing Section described in Chapter 3 z Use Cases implemented 1 ARVAnalysis 2 Save HRVAnalysis 5 6 Control Component Architecture 51 HRVAnalysis Required Initialized components ItfConHRVAnalyss O itmtoane Created Components Figure 5 11 Component Diagram of HRV Analysisl control 5 6 7 Utilities 1 Brief Description This component is responsible for allowing database utilities to be requested and performed The only two options alowed are database compression and password changing Use Cases implemented 1 Change Password 2 Compact Database 52 CHAPTER 5 Architecture Required Initialized components ItfConMenuPrincipsl Created Components Figure 5 12 Component Diagram of Utilities control In the next chapter The enxt chapter will present the results and depict the state of the art of the application compared to other software Chapter 6 Discussion Don t fear failure so mu
2. B SAVE amp NEXT S RESET FIELDS SAVE amp RR ANALYSIS e MAIN MENU a ka Jun 18 2013 El Palabra 1 Mitral valve prolapse None Current Conditions Smoker currently Caffeine on past 24h Alcohol on past 12 h Exercise over past 24 h Figure 4 2 Insert ECG Frame In this window we will be able to associate all relevant information of the patient to the ECG entry without giving away any important demographic details This prevents any intruder if he was able to decrypt and access database data to match any of the recorded information to the physical patient T he options we thought that were relevant are shown in the window None of the options are mandatory except for the data of the itself add new ECG the import process must be successful Below more detailed explanation of the meaning for each field is shown Gender If the patient is Male or Female Age Age of the patient when the ECG was recorded If it is unknown set it to 1 Date Date the ECG was recorded Its default value is the current day If it is not known you can click on it and then select none Heart Failure Select the value if the patient has suffered of a heart 20 CHAPTER 4 ECG2HRV User Guide failure following NYHA New York Heart Association Functional Clas sification Classification z Mitral Valve Prolap
3. Table 3 1 Selected time domain measures for HRV from Camm et al 1996 associated with a duration These values are not equally distributed along time axis That is the reason to need to re sampling data at a rate that will not miss any beat For the re sampling process we used a linear interpolation approach with a sampling rate of 4 Hz Berger et al 1986 The fact that it is physiologically impossible to have a beat with length smaller than than 0 25 seconds ensures that there is no NN interval value missed The re sampling values depend on which NN intervals are comprised be tween its current values and the next If all the NN interval is inside that 500 ms time span then the value of the NN interval is taken On the other hand if the time span contains two segments of different NN intervals a linear interpolation with a window size of 500 ms is made to calculate each re sampling point Figure 3 2 shows an example of how the process works obtained from Berger et al 1986 Now we have an evenly time spaced signal whose spectrum can be easily recognized by using FT Fourier Transform For this purpose we decided to compute the FT by using a FFT Fast Fourier Transform algorithm implemented by the library Eichelberger y li 2002 The power spectrum is extracted from the FFT output and added to get the parameters displayed on Table 3 2 14 CHAPTER 3 Signal Processing 1 12 4 143 14 4 b t
4. 47 CONTENT TABLE XIII 5 6 4 Search EGG ius eU EIUS edu Ea 5 65 RR Detection vr 5 6 6 HRV Analysis 2252 rox iki 22446 72 62 5 6 7 Utilities In the next chapter 6 Discussion In the next chapter References Acronym List 48 49 50 51 52 53 55 57 59 Figure Index 2 1 Anatomic structure of the heart QRS compex 6 2 2 Example of RR interval 7 3 1 Signal proccessing steps for ECG segment 11 3 2 REsampling algorithm 14 4 1 Login Dialog Frame 18 4 2 Insert ECG Frame 19 4 3 Search ECG 21 4 4 RR detection Frame 24 4 5 HRV Analysis 25 4 6 Utilities Frame 2 4 Lacs oem ROB RH 26 4 7 Change Password 26 5 1 Global Use Case 29 5 2 Class Domain First 39 5 3 Class Domain second part 40 5 4 Database Schematic Diagram 43 5 5 Visualization Component 45 5 6 Component Diagram of Login control 46 5 7 Component Diagram of Main Menu control 47 5 8 Component Diagram of InsertECG 48 5 9 Component Diagram of SearchECG control 49 5 10 Component Diagram of R
5. HUIKURI H Heart rate variability in coronary artery disease Journal of internal medicine vol 237 4 p ginas 349 357 1995 NISKANEN J P TARVAINEN M P RANTA AHO P O y KARJALAINEN P A Software for advanced HRV analysis Computer Methods and Programs in Biomedicine vol 76 1 p ginas 73 81 2004 ISSN 0169 2607 PAGANI M Heart rate variability and autonomic diabetic neuropathy Di abetes nutrition amp metabolism vol 13 6 p gina 341 2000 PAN J y TOMPKINS W J A real time qrs detection algorithm Biomedical Engineering IEEE Transactions vol 3 p ginas 230 236 1985 POMERANZ B MACAULAY R CAUDILL M A KUTZ ADAM D GORDON D KILBORN K M BARGER A C SHANNON D COHEN R J ET AL Assessment of autonomic function in humans by heart rate spectral analysis American Journal of Physiology Heart and Circulatory Physiology vol 248 1 p ginas H151 H153 1985 TARVAINEN M P NISKANEN J P LIPPONEN J RANTA AHO P y KARJALAINEN P Kubios hrv a software for advanced heart rate variabil ity analysis En 4th European Conference of the International Federation for Medical and Biological Engineering p ginas 1022 1025 Springer 2009 VOLTERRANI M SCALVINI S MAZZUERO G LANFRANCHI P COLOMBO R CLARK A L y LEVI Decreased heart rate vari ability in patients with chronic obstructive pulmonary disease CHEST Journal vol 106 5 p ginas 1432
6. y SILVERI Heart rate variability and my ocardial infarction systematic literature review and metanalysis Eur Rev Med Pharmacol Sci vol 13 4 p ginas 299 30 2009 CAMM A J MALIK M BIGGER J BREITHARDT G CERUTTI S COHEN R COUMEL P FALLEN E KENNEDY H KLEIGER R ET AL Heart rate variability standards of measurement physiological inter pretation and clinical use task force of the european society of cardiology and the north american society of pacing and electrophysiology Circula tion vol 93 5 p ginas 1043 1065 1996 EICHELBERGER H y Ir L Jtransform a java source code transforma tion framework Informe t cnico TR 303 Institute for Computer Science Wurzburg University 2002 EINTHOVEN W The string galvanometer and the measurement of the action currents of the heart Nobel Lecture December vol 11 1925 GEIGER A J Uses and limitations of electrocardiography in the diagnosis of acute coronary occlusion With particular reference to the latency of changes The Yale Journal of Biology and Medicine vol 11 6 p gina 619 1939 GOLDBERGER L AMARAL L GLASS L HAUSDORFF J M IvaNov P C MARK R G MIETUS J E MooDY B PENG 97 58 REFERENCES C K y STANLEY H E Physiobank physiotoolkit and physionet Com ponents of a new research resource for complex physiologic signals Cir culation vol 101 23 p ginas 215 220 2000
7. 2003 The application from Na tional Instruments has a rich visual and easy to use interface with comprehensive on line tutorials The shortcoming is that it does not allow to store ECGs nor to associate them with patient background Our application may not have some of the functionality offered here like SDNN graph correlation yet but our data structure makes ECG2HRV more powerful and comfortable for the user by giving away some pa rameter analysis in exchange of patient background association It is also subject to be used by buying the required software while our so lution is free and open source Software for advanced HRV analysis Niskanen et al 2004 This C application includes visual libraries and has same functionality as XU et al 2003 It lacks from the same database characteristic explained before and may not be compiled to work an different operating systems Kubios HRV Tarvainen et al 2009 This software is an easy to use HRV report generator It is useful to get isolated ECG reports but not to be used as daily basis It works under Linux GNU and Windows but not under Mac computers Many other applications exist to satisfy HRV calculation and reporting requirements but none to relate obtained results to patient background We have conceived designed and implemented an ECG data management tool that permits automatic analysis of HRV allowing modification of RR inter val detections due to ectopic beats noise
8. IttConMenuPrincipsl amp e principal D itconRRDetection ItfConMenuPrincipal ItfModDB k da O tWisinsertarECG MiConDefacctonRR aj 1tFModDB Created Components LitfConBuscarECG LL QD ieonRRDetection Figure 5 8 Component Diagram of InsertECG control 5 6 4 Search ECG Brief Description This component is responsible for giving the user the option to SearchECG and any related ECG selection action Use Cases implemented SearchECG ModifyECG just the selection and request DeleteECG ModifyECG just the selection and request HRVAnalysis just the selection and request Ct A W N DetectRR just the selection and request 5 6 Control Component Architecture 49 BuscarECG 7 Required Initialized components ItfConBuscarECG 9 Created Components EN Figure 5 9 Component Diagram of SearchECG control 5 6 5 RR Detection 1 Brief Description This component is responsible for performing the RR Interval detection and enabling its modification through a visual interface Use Cases implemented DetectRR Save RR detection Modify RR detection HRVAnalysis just the request B WO N e 50 CHAPTER 5 Architecture RRDetection 7 Required Initialized components Created Components
9. double normal List lt Boolean gt time double to List lt Integer gt value double getDuration List lt Double gt ECGData double double double getFrom List lt Integer gt ECGData getNormal List lt Boolean gt getTime double getNormalDurations double getValue double getTo List lt Integer gt isEmpty boolean e getTotalCorrectDuration double loadObject InputStream void loadObject InputStream void recalculateTime void RRDetection saveObject String void RRDetection List lt Integer gt List lt Integer gt List lt Double gt List lt Boolean gt setTime double void saveObject String void setValue double void setDuration List lt Double gt void setValues double double void setFrom List Integer void property get setNormal List Boolean void getFs double setTo List lt Integer gt void property set setFs double void dati rrintervals ECG Serializable data ECGData HRVAnalysys ECGInfo frequency double HF double HFNorm double LF double LFHFRatio double LFNorm double NN50Count int pNN50 double RMSSD double SDNN double SDSD double spectrum double VLF double hRVAnalysis HRVAnalysys Keywords List lt KeyWord gt rrintervals RRDetection ECG ECGInfo getData EC
10. Definitions eS MUERTA UNA TR 27 5 2 1 Cage Y x GSE WoO aoe thy Se SE sa 27 5 2 2 Class 28 5 23 Architecture 28 5 2 4 Component 28 5 2 5 Use Case Diagram 28 5 2 6 Model View Controller 28 5 3 Use Case Document 28 Darle MOB b ONA A AY 29 A lee a cec A 30 5 3 3 Search ECG R 31 5 34 Insert 31 5 3 5 Modify ECG ss k big dus gebe Reb 4 32 5 3 6 33 5 3 0 Import ECG 2 eel i RD 33 5 83 82 ia nto Ee E MSS S 34 5 8 9 Modify 34 5 3 10 Save RR detection 35 5 3 11 HRV Analysis 36 5 3 12 Save HRV 36 5 3 13 Change 37 5 3 14 Compact Database 38 5 4 Data Model dne AA eee ALTOS 38 5 4 1 Class Description ee 38 5 4 2 Database implementation 42 5 5 Visual Component 42 5 6 Control Component Architecture 45 5 61 A OU MON o NATA ONT 45 5 6 2 Mali Ment 2 uu vem momo oto s ees 46 5 6 3 Insert
11. Society of Pacing and Electrophysiology standarized the 5 minutes and 24 hours ECG recording markers for HRV in order to prevent incorrect conclusions from excessive or unfounded extrapo lations of such measures There are several domains when it comes to HRV each of them with its own marker In this project we only consider 5 minutes ECG recordings The time domain markers are shown in Table 3 1 and the frequency domain domain markers are at Table 3 2 In the next chapter Once we have explained the required concepts we will proceed to explain how we filter the signal to remove artifact process it to detect R peaks and describe the transformations required to perform PSD Power Spectral Density of the series of NN intervals to obtain all HRV parameters Chapter 3 Signal Processing The process of scientific discovery 18 im effect a continual flight from wonder Albert Einstein ABSTRACT This chapter describes the process that was developed in order to get HRV analysis from a single lead ECG raw signal at any sampling rate 3 1 Process overview This procedure can be divided into the following subsequent tasks Filtering the signal to remove artifact Preprocessing the filtered signal for the next step Detecting NN intervals disregarding abnormal ones Convert irregular time function of intervals to a regular continuous output Calculate time domain HRV parameters Calculate frequency domain
12. correct input and validates the user 30 CHAPTER 5 Architecture 5 3 2 3 The application start the next component to show program op tions and closes the login dialog Alternate event flow enumeration 1 The application starts and shows an input dialog 2 The application receives wrong input and validates the user 3 If it is not the third try the application will let the user validate again if it is the third program will terminate execution Special requirements the user must have Java Virtual Machine in stalled in order to launch Preconditions User must know the password Postconditions If successful the actor User will mutate to Validated User and start the next module and the database model interface Exit Brief Description this Use Case describes how to exit and what to do depending upon the actor that requests this feature Event flow ennumeration 1 User requests exit 2 Program closes visualization and terminates Alternate event flow enumeration 1 Validated User requests exit 2 Database gets shut down 3 Program closes visualization and terminates Special requirements None Preconditions Application must be running Postconditions Application will shut down memory will be freed database will be safely closed and all processes related to ECG2HRV will be terminated 5 3 Use Case Document 31 5 3 3 Search ECG a Brief Description This Use Case may be the most impo
13. observations patients with reduced high frequency HRV improved their numbers by more than ten times when being tested while using those devices Paul had a good guess and it took him around twenty years of field work to do it so Can you imagine if he could check if that hypothesis was correct with past twenty year patient data What is more can you think of how easy it could be for him to perform a simple search in order to discover if he may be right by comparing all the data he had from all his patients This functionality may not only lead him to check if an hypothesis may be correct but to discover new possible research lines by isolating patient groups with certain characteristics associated to each ECG ECG2HRV is intended to promote research in Heart Rate Variability by offering the user required tools to store process and search HRV for a raw ECG detaching the user from signal processing If Paul had started using this tool twenty years ago he would probably had arrive to that intuition or another way before than he did 1 5 Chapter Structure This document is structured with the following sections CHAPTER 1 Introduction Chapter 2 will define the required concepts to understand what this application does and the physiological proporties that lay behind ECG Chapter 3 explains in detail the algorithms developed to get the industry standard HRV measurements along with all the process description Chapter 4 spans the e
14. or miss detections This software is platform independent so same program is able to run under Windows Linux GNU Mac Solaris or any operating system that is able to run Ja va Virtual Machine in a desktop platform The fact of being open source enables the community to carry on with improvements and additional tools that could expand its user impact This condition also makes the applica tion perfect for low budget research give that it is free to use We hope that ECG2HRV does not stop growing and that the community will continue improving its available features In the next 55 In the next chapter Like every day has a night or every yin has a yang every beginning has an end This chapter finalizes the documentation for the project ECG2HRV We hope that its contents have helped you to understand how it works how to use it and how to improve it The next sections will show the biblio graphic references used the appendix and a list containing all the acronyms employed References Y as del mucho leer y del poco dormir se le sec el celebro de manera que vino perder el juicio Miguel de Cervantes Saavedra BERGER R D AKSELROD 5 GORDON D y COHEN R J effi cient algorithm for spectral analysis of heart rate variability Biomedical Engineering IEEE Transactions vol 9 p ginas 900 904 1986 BUCCELLETTI E GILARDI E SCAINI E LEONARDAGALIUTO PER SIANI R BIONDI
15. relate any new ECG to that keyword Removing any keyword from the database is possible but doing so will also delete any relationship between any other ECG to that keyword import an ECG you need to have the data a CSV Comma Sepa rated Values file using the following format each value will represent the voltage output for the ECG at the specified time position The time will 4 4 21 be recreated with the sampling frequency Fs Once the data follows the restricted format click on the folder button and select the file location ITs contents will be loaded and ready to be stored at the database When all the desired inputs are filled and the ECG has been succesfully loaded the user will be able to save and insert a new one or to detect the RR intervals in order to perform the HRV analysis If the user decides not to perform RR analysis this option will be available selecting in the search mode the ECG and clicking on RR analysis Note that a specific ECG can be hard to distinguish from others except for its identification number so we strongly reccomend detecting the RR intervals and subsequent HRV analysis right after the ECG insertion In this way HRV parameter will be stored along with the ECG data and related information There two more options we have not mentiones yet Reset fields and Main menu The first one will roll back all fields to its default value while the second one will ret
16. setDatelmported Date void setDiabeticNeuropathy boolean void setExercisePast24h boolean void setHeartFailure HeartFailure void setHypertension boolean void setld int void setMale boolean void setMitralValveProlapse MitralValveProlapse void setRecentMyocardiacinfarction boolean void setRecurrentVasoVagalSincopalEpisodes boolean void setSinusRhythm boolean void setSmoker boolean void setUntreatedDepression boolean void info ECGSearch ageFromEnabled boolean false ageTo int ageToEnabled boolean false alcoholPasdt12hEnabled boolean false analysisFrom HRVAnalysys analysisTo HRVAnalysys caffeinePast24hEnabled boolean false cardiomyopathyEnabled boolean false datelmportedFromEnabled boolean false datelmportedTo Date datelmportedToEnabled boolean false diabeticNeuropathyEnabled boolean false exercisePast24hEnabled boolean false heartFailureEnabled boolean false HFFromEnabled boolean false HFNFromEnabled boolean false HFNToEnabled boolean false HFToEnabled boolean false hypertensionEnabled boolean false info ECGlnfo LFFromEnabled boolean false LFHFRatioFromEnabled boolean false LFHFRatioToEnabled boolean false LFNFromEnabled boolean false LFNToEnabled boolean false LFToEnabled boolean false maleEnabled boolean false mitralValveProlapseEnabled boolean false NN50CountFromEnabled boolean false NN50CountTo
17. the clinician as well as the ECG itself providing an oppor tunity for the user to override or exclude any part of the ECG determined not acceptable for the analysis This mobile device based tool is useful be cause it supports a new paradigm in clinical diagnosis that joins the power of computer and database analysis helping to work out a differential diagnosis for each patient 1 3 Who is intended to use ECG2HRV If you are a physician or researcher that deals everyday with 5 minutes ECG info this may be the perfect tool for you It enables you to safely store all the collected data while maintaining its order ECG2HRV should not only be used as a data management tool but also as a source of research ideas or for research itself In built ECG2HRV capacity to automatically detect all the NN intervals allows to calculate all the HRV parameters with a single click instantly translating them to a time gain for the user No more complex programs or software will be needed to get the desired HRV markers In addition to this comfortable advantage this software allows to perform different searchs in which any field can be crossed with each other so that ECG groups can be isolated to identify potential research lines or to select a particular block of patients The only requirement this software needs is having a reliable source of ECG measurement Technology is evolving in this field and many devices are being approved by the FDA US Food a
18. the user to delete it Postconditions If successful deletes the selected ECG from the pre sistence 5 3 7 Import ECG Brief Description This characteristic will be responsible of import ing ECG data from the selected source a Event flow enumeration 1 Validated User selects ECG file data location from any operating system drive disk and inputs sampling rate 2 program verifies the format of the file and imports its data Alternate event flow enumeration 1 Validated User selects ECG file data location from any operating system drive disk 2 program fails to validate file format 3 An error message is shown to Validated User 34 CHAPTER 5 Architecture Special requirements Being Validated User Preconditions Being inserting or modifying an ECG record Having a CVS file with the required format this is sampled voltage values separated by commas and the knowledge of the recording sampling rate Postconditions If successful adds data content from the file to the current ECG model Data will not be stored until save Use Case is selected 5 3 8 DetectRR Brief Description This functionality will enable automatic detection of RR Intervals and its classifications as normal NN or not normal Event flow enumeration 1 Validated User requests RR interval detection for a particular ECG from Search or new Insert 2 The program checks that ECG has no RR Interval detecti
19. 1437 1994 Wu WANG C J YANG Y K WANG J S y CHUNG P C Emotional quality level recognition based on hrv Neural Networks IJ CNN The 2010 International Joint Conference on p ginas 1 6 IEEE 2010 XU X T LU G w YI W J y ZHOU W B The application of labview techniques in hrv research j Chinese Journal of Medical Physics vol 2 p gina 020 2003 Acronym List CHE Congestive Heart Failure COPD Chronic Obstructive Pulmonary Disease OSV russia Comma Separated Values ECG Electrocardiogram ECG2HRV Electrocardiogram to Heart Rate Variability FDA US Food and Drug Administration EET Fast Fourier Transform Moe Fourier Transform HRV Heart Rate Variability NYHA New York Heart Association Functional Classification PSD i26 Power Spectral Density Unified Modeling Language 59 60 REFERENCES Autorizo a la Universidad Complutense a difundir y utilizar con fines acad micos no comerciales y mencionando expresamente a sus autores tan to la propia memoria como el c digo los contenidos audiovisuales incluso si incluyen im genes de los autores la documentaci n y o el prototipo desa rrollado Fdo Miguel ngel Men rguez Garc a
20. ECG2HRV NEW TOOL FOR THE RESEARCH IN HEART RATE VARIABILITY PROYECTO DE SISTEMAS INFORM TICOS Miguel ngel Men rguez Garc a Facultad de Inform tica Universidad Complutense de Madrid Julio 2013 ECG2HRV NEW TOOL FOR THE RESEARCH IN HEART RATE VARIABILITY ECG2HRV Electrocardiogram to Heart Rate Variability Project report IT 2013 07 Supervised by Drs Jose Ignacio Hidalgo Universidad Complutense de Madrid John Dyer University of Oklahoma Facultad de Inform tica Universidad Complutense de Madrid Julio 2013 Copyright Miguel ngel Men rguez Garc a Special Regards En el estudio no existe la saciedad Erasmo de Rotterdam 1469 1536 Intelligence is the ability of sorting and interlacing all acquired knowledge and possible resources in order to efficiently solve problems that may arise during our lives Education is the cornerstone of such process each learned concept entails a new tool to solve such matters Firstly I want to aknowledge all the time effort and support that my parents have brought me during these past years In second place I would like to thank the Complutense University of Madrid UCM for offering me the opportunity to improve the tools I am able to use today as well as to bring me the chance to learn similar contents from different points of view thanks to study abroad scholarships The year I studied at Bordeaux France made my curiosity bloom to learn fr
21. Enabled boolean false noHRVAnalysis boolean false PNNSOFromEnabled boolean false PNNSOToEnabled boolean false recentMyocardiaclnfarctionEnabled boolean false recurrentVasoVagalSincopalEpisodesEnabled boolean false RMSSDFromEnabled boolean false RMSSToEnabled boolean false SDNNFromEnabled boolean false SDNNToEnabled boolean false SDSDFromEnabled boolean false SDSDToEnabled boolean false sinusRhythmEnabled boolean false smokerEnabled boolean false untreatedDepressionEnabled boolean false VLFFromEnabled boolean false VLFToEnabled boolean false words List lt KeyWord gt ChangePassword newP char 0 oldP char ChangePassword char char mitralValveProlapse MitralValveProlapse id int value String equals Object boolean getld int getValue String hashCode int MitralValveProlapse int String setld int void setValue String void toString String ttt b Gk B GB heartFailure gt gt getNewP char getOldP char HeartFailure id int value String tt ORO ROB GR equals Object boolean getld int getValue String hashCode int HeartFailure int String setld int void setValue String void toString String Figure 5 3 Class Domain second part ECGSearch does not show a
22. GData getEcglnfo ECGInfo getKeywords List lt KeyWord gt getRrintervals RRDetection setData ECGData void setEcglnfo ECGInfo void setkeywords List lt KeyWord gt void setRrintervals RRDetection void setValues Double Double boolean tt tt tt tt getFrequency double getSpectrum double HRVAnalysys setFrequency double void setSpectrum double void tet tee property get getHF double getHFNorm double getLF double getLFHFRatio double getLFNorm double getNNSOCount int getpNN50 double getRMSSD double getSDNN double getSDSD double getVLF double property set setHF double void setHFNorm double void setLF double void setLFHFRatio double void setLFNorm double void setNN50Count int void setpNN50 double void SetRMSSD double void setSDNN double void setSDSD double void setVLF double void OROGROR ORB GB GR ORGR OR GB GB R6 6 HRVAnalysys double double double int double double double double double double double property get gethRVAnalysis HRVAnalysys property set sethRVAnalysis HRVAnalysys void hRVAnalysis ECGSearch K analysisFrom ECGSearch E analysisTo KeyWord description String dd int value String equals Object boolean getDescription String getld int getV
23. HRV parameters 3 2 Filtering Process ECG signal can be misinterpreted due to different kind of artifacts Re moving them to get a clearer signal is vital to get neat results The main artifact sources of ECG are 10 CHAPTER 3 Signal Processing interference Alternating Current is the power source used for all electronic equipment can cause the signal to look like thick and hairy z Baseline Wonder When the isoelectric line changes its position a roller coaster like behavior is observer in the output changing the level that is considered to be the ground This interference usually is the low frequency band of the ECG Muscle artifact Movement of the muscles generate electric impulses that are recorded by the leads and can obscure ECG output When muscle tremors occurs continuous random values are passed to the leads making sometimes the signal unintelligible To remove most of the artifacts described above we followed part of the process proposed at Pan y Tompkins 1985 with some modifications They propossed a bandpass filter composed by a high pass low pass sequential filtering process Instead we used a butterworth filter of 6th order to allow only frequencies in the range of 2 20 Hz This procedure gets rid of two artifacts baseline wondering dissapears because it is below the low cut frequency and AC interference of 50 or 60 HZ gets removed too from the output because of the fact that it exceed he
24. HRV report for current NN intervals Rollback RR intervals will be reseted to las stored value on the database a Back Returns to the search frame graph under the top panel allows visualizing ECG s RR and NN intervals overlaying its signal using red and green rectangles with its duration as height value Modification of the beginning and end of the intervals is allowed by clicking on the desired R peak and dragging it to the new position If the dragging process ends near another RR interval the selected peak will disappear Clicking on a segment far enough from any R peak will create a new R peak detection where the user releases the mouse if it is in between the previous and the next R peak and far enough from both After doing any changes do not forget to press on save in order to per manently store the modified detections 4 6 HRV analysis Once the user has logged in he will be able to search any of the inserted ECGs by selecting Search ECG option in the main menu then a window like the one shown in Figure 4 5 will be displayed 24 CHAPTER 4 ECG2HRV User Guide RR interval detector QJ PREV MISSDETECT gt NEXT MISSDETECT HRV REPORT 4 SAVE CHANGES 4 RESET RR DET x ROLLBACK 1 0 Time s 87 0 87 5 88 0 88 5 89 0 89 5 90 0 90 5 91 0 91 5 92 0 92 5 93 0 93 5 94 0 94 5 95 0 95 5 96 0 96 5 97 0 Voltage mv Fe gt Figure 4 4 RR detection Frame HRV Analysis
25. RDetection control 50 5 11 Component Diagram of HRV Analysis control 51 5 12 Component Diagram of Login control 52 XV Table Index 3 1 Selected time domain measures for HRV from Camm et al 1996 13 3 2 Selected frequency domain measures for HRV from Camm et al 1906 5 A 14 Chapter 1 Introduction ABSTRACT This chapter will introduce the importance of Heart Rate Variability and depict the structure of this document 1 1 Introduction ECG is well known to be a marker for various pathologies of the cardiac excitation cycle When the ECG is collected from multiple leads in the clas sic 12 lead arrangement a number of mechanical heart pathologies can also be detected and diagnosed Geiger 1939 Einthoven 1925 More recent ly HRV is a secondary parameter that has been extracted from the ECG which helps differentiate the clinical diagnosis of heart and lung diseases such as COPD Chronic Obstructive Pulmonary Disease and CHF Con gestive Heart Failure Volterrani et al 1994 Huikuri 1995 HRV is closely related to the activity of the autonomic branch of the body s central ner vous system Pomeranz et al 1985 and is also predictive of potential heart rhythm pathologies in elderly patients ECG measurement devices have built in implementation of the HRV pa rameters but those values are difficult to extract and relate
26. alue String hashCode int KeyWord int String String setDescription String void setld int void setValue String void toString String tt et Figure 5 2 Class Domain First part 40 CHAPTER 5 Architecture ECGlnfo age int alcoholPasdt12h boolean caffeinePast24h boolean cardiomyopathy boolean datelmported Date diabeticNeuropathy boolean exercisePast24h boolean heartFailure HeartFailure hypertension boolean id int male boolean mitralValveProlapse MitralValveProlapse recentMyocardiaclnfarction boolean recurrentVasoVagalSincopalEpisodes boolean sinusRhythm boolean smoker boolean untreatedDepression boolean ECG ecginto gt tt b bt bb bt tt tb FFF RR RR RR ECGInfo getAge int getDatelmported Date getHeartFailure HeartFailure getld int getMitralValveProlapse MitralValveProlapse isAlcoholPasdt12h boolean isCaffeinePast24h boolean isCardiomyopathy boolean isDiabeticNeuropathy boolean isExercisePast24h boolean isHypertension boolean isMale boolean isRecentMyocardiaclnfarction boolean isRecurrentVasoVagalSincopalEpisodes boolean isSinusRhythm boolean isSmoker boolean isUntreatedDepression boolean setAge int void setAlcoholPasdt12h boolean void setCaffeinePast24h boolean void setCardiomyopathy boolean void
27. ch that you refuse to try new things The saddest summary of a life contains three descriptions could have might have and should have Louis E Boone ABSTRACT This section will summarize what objectives we have obtained with ECG2HRV and suggest future lines of work that could improve what has being developed ECG2HRV brings a new paradigm in HRV research field that may seed new ideas for time and frequency domain parameter correlation for clinical diagnosis This feature reliefs the user from the heavy task of carrying about storage format while enabling same functionalities other programs have in a single binary This is not the first application that is conceived to calculate HRV A web survey conducted reveals that ECG2HRV is the only one that allows ECG information to be related to HRV analysis while offering auto matic calculation of required information within a single interface Next we will discuss some examples supporting these statements HRV Toolkit Goldberger et al 2000 The application from Physionet is composed by a set of C files that permit obtaining HRV analysis using the command line in time and frequency domain from MIT BIH database files Its interface is complicated to handle and many potential users are discouraged by this fact If we add the point that it does not offer any system to store ECGs we realize that ECG2HRV is far more useful 53 54 CHAPTER 6 Discussion LabView HRV Analysis XU et al
28. d Signal The second part of this process consists in creating an integrated moving average window The size of the window has to be wide enough to create hill like waves as shown in Figure 2 c but not as big as to merge R peak and T waves A default size of 0 2 F Sampling Frequency as suggested in ensures optimality based on all the empirical tests that were made results for this processing are shown for a three QRS complex example illustrated in Figure 3 1 c last step of the preprocessing part is to derivate the output of the moving average window single and not computationally expensive algorith was used based on formula 3 2 The purpose of this last part is to detect the rising and the falling edge of the hill so that pattern can be recognized in the following step The output of this part is shown in Figure 3 1 d yin 1 yn 3 2 4 b A N PA 1 ok AJ ART 1 mi A Al E W ur ui Ny WIN Figure 3 1 Preprocessing Original ECG b ECG after bandpass filter c Bandpass filter output after applying the moving integration window d Differ entiation from the moving integration window 12 CHAPTER 3 Signal Processing 3 4 NN Interval Detection Differentiated signal allowed us to recognize the QRS complex with the unique feature described above but i
29. d for the search The rest of boolean variables are indicators of whether the represented value is going to be used for the search or not 1 HeartFailure Class representing the different values of heart failures following NYHA convention a HRVAnalysis This class will store the results of the Heart Rate Vari ability analysis in time and frequency domain a KeyWord Represents the associations the doctor want to include for and ECG but is not able to because it is not included by default in its properties a MitralValveProlapse Object representing the different stages of Mi tral Valve Prolapse a RRDetection This class contains the information required to recre ate the RR Interval detection done by the application and probably modified by the user As in the case of ECGData this class imple ments serializable for efficiency reasons 42 CHAPTER 5 Architecture 5 4 2 Database implementation Relational databases offer exceptional characteristics for data storage in cluding security information coherency ease of use and many other proper ties We decided to use this paradigm so that searches could be execut ed rapidly and to avoid dealing with information corruption Relational database operations are usually subject to good properties such as ACID which stands for Atomicity Consistency Isolation and Durability Between all the available options we chose Apache Derby database for its embeddable simplicity which fre
30. d it may improve its integrity In the next chapter This chapter may be useful for users of the application but developers require much more information than a simple guide The next chapter will explain he architecture of the application so that the coder is willing to modify this piece of software byto adding a new feature or improving it he won t face to a hard task 26 CHAPTER 4 ECG2HRV User Guide UTILITIES ECG2HRV Electrocardiogram to Heart Rate Variability CHANGE PASSWORD 2 COMPACT DB ge BACK Figure 4 6 Utilities Frame CHANGE PASSWORD C3 Change Password Dialog vA OK X CANCEL Figure 4 7 Change Password Frame Chapter 5 Architecture Structure is one of the things that I always hope will reveal itself to me Richard Russo ABSTRACT This section will describe the architecture the application has followed and its relations following UML notation and standard diagrams 5 1 Overview This chapter is not intended to explain how the project was developed but to detail its functionalities and inner structure The purpose is to relieve the difficulty of modifying this program for unknown coders First we will introduce the definitions of several concepts that will be used during this section Then the Use Case section will describe each of the implemented functionalities at a high concept level Next section will deal w
31. d the process in which the HRV param eters is calculated so that the user can verify that the output ECG2HRV is giving 18 correct The next chapter contains a guide detailing the program functions with visual reference so that the end user can check if he has any questions on its use Chapter 4 ECG2HRV User Guide It is my fervent wish and my greatest ambition to leave a work with a few useful instructions for the pianists after me Franz Liszt ABSTRACT This chapter will present ECG2HRV different function alities and will detail those features This section is intended to be read by new users or individuals with questions about a particular characteristic of the application 4 1 System Requirements is a desktop cross platform application that can run in dif ferent software architectures like Windows Mac an any Linux GNU dis tribution if they have JAVAO virtual machine installed This feature frees the user from being forced to use a particular operating sysem which may be unfamiliar for him This application is not intended to run on mobile phones and it will probably not work properly on any of those even if they are JAVAO enabled The visual interface was build with a default desktop screen size If your operating system does not have JAVA installed the required files can be downloaded and isntalled from Oracle webpage To install ECG2HRV simply download the zip file and extract its content to th
32. d to use ECG2HRV 2 1 4 History example of a potential ECG2HRV user 3 1 5 Chapter Structure 3 2 Background 5 2 1 What is an ECG Electrocardiogram 2 5 2 2 What is an RR 6 2 3 What is HRV Heart Rate Variability 7 In themext chapter ers 7 3 Signal Processing 9 3 1 Process overview 9 3 2 Filtering Process ta eem ERES 9 3 3 Preprocessing Filtered Signal 11 3 4 NN Interval Detection 12 3 5 Time Domain HRV Parameters 12 3 6 Frequency HRV Parameters 12 In the next 15 4 ECG2HRV User Guide 17 4 1 System 17 4 2 Rn a Meee dean eae Vut 18 XII CONTENT TABLE 43 ECG Insertion 18 4A WOG Search oz Rx Xx Ru d 21 4 5 RRs Detection aea eC A Aix NANI 23 1 6 HRV Analysis Revo 23 Ass A unen tox Ne RII RM ws ee 24 4 7 1 Change 24 4 7 2 Compact database 24 In themext chapter a E A 25 5 Architecture 27 Dl OVerview da 27 522
33. dated with the average value of the last eight positive R peaks A simple check is used to detect if it is a valid candidate for R peak if the current valley exceeds k THRESHOLDS3 then the algorithm checks that possible R peak if not it skips that value 3 5 Time Domain HRV Parameters The selected time domain HRV parameters are displayed in Table 3 1 values are calculated from the NN interval output explained before 3 6 Frequency HRV Parameters obtain frequency domain parameters a more complicated process is required The NN interval function is composed of a succession of time values 3 6 Frequency HRV Parameters 13 Variable Units Statistical measures SDNN ms Standard deviation of all NN intervals SDANN ms Standard deviation of the averages of NN intervals in all 5 min segments of the entire recording RMSSD ms square root of the mean of the sum of the squares of differences between adjacent NN inter vals SDNN index ms Mean of the standard deviations of all NN intervals for all 5 min segments of the entire recording SDSD ms Standard deviation of differences between adjacent NN intervals NN50 count ms Number of pairs of adjacent NN intervals differing by more than 50 ms in the entire recording Three variants are possible counting all such NN intervals pairs or only pairs in which the first or the second interval is longer pNN50 NN50 count divided by the total number of all NN intervals
34. denador personal Este software permite al usuario analizar autom ticamente resultados de un ECG parar averiguar la VFC y almacenar la informaci n en una base de datos junto con todos los datos no demogr ficos relevantes del paciente Para ello hemos desarrollado un nuevo algoritmo de detecci n de los complejos QRS de un ECG de baja complejidad que permite al usuario obtener un an lisis de la VFC en el do minio del tiempo y de la frecuencia Para efectuar el an lisis del espectro de la frecuencia se ha implementado un algoritmo prpuesto por Berger et al 1986 que muestrea a 4 Hz la se al irregular en el tiempo transform ndola en continua para poder realizar an lisis de Fourier El proceso es transpar ente para el usuario de tal forma que muestra nicamente la detecci n de intervalos RR y el informe de la VFC ahorrando de esta forma tiempo al in vestigador Esto permite tambi n el almacenamiento de los resultados junto con toda la informaci n relevante del ECG con el fin de que pueda ser uti lizada en b squedas cruzadas de pacientes con determinadas caracter sticas cl nicas RESUMEN Palabras Clave HRV ECG EKG VFC ECG2HRV VARIABILIDAD FRECUENCIA CARD ACA BBDD ANALISIS Contents Special Regards v Abstract VII Resumen 1 Introduction Til 1 2 What is ECG2HRV Electrocardiogram to Heart Rate Vari ability JI b dude hordei m esu ots ae Oe ere 2 1 3 Who is intende
35. e desired isntallation folder Do not touch any of the contents of that folder If desired a shortcut can be created to ECG2HRV jar 17 18 CHAPTER 4 2 User Guide Regarding the execution of the program right click on ECG2HRV or its shortcut and right click it then select your installed JVM Java Virtual Machine After a few seconds the program will start 4 2 First Run The first time the program is executed will be like the rest Run the program as described in 4 1 a window like the picture shown in Figure 4 1 will appear To access in the Login dialog introduce the default password myPassword We strongly reccomend selecting Utilities in the Main Menu and the click on change password option in order to prevent malicius users fron accessing any of the future contents that the database will have There is nothing more that is required to be done differently than the rest of the runs Now you are ready to start ECG insertion LOGIN WINDOW ECG2HRV Electrocardiogram to Heart Rate Variability Cancel JP Validate Figure 4 1 Login Dialog Frame 4 3 ECG Insertion Once the user has logged in he will be able to insert new ECGs to the database by selecting Insert ECG option in the main menu Then a screen 4 3 ECG Insertion 19 like depicted in Figure 4 2 will be shown INSERT NEW ECG Add new ECG D to Heart Rate Variability
36. eeds They are composed by the following objects Communication This class is the link between the graphic content and control Any important operation must be called by using this object functions We would like to remark that this class does not 5 5 Visual Component Architecture 43 IDHEARTFAILURE O IDHEARTFAILURE VALUE P IDECG GENDER SMOKER ALCOHOL12 EXERCISE24 CARDIOMIOPATHY HYPERTENSION VASOVAGAL DEPRESSION MYOINFARCTION IDHEARTFAILURE DAY IDMVP NEUROPATHY FREQSAMPLING ECG AGE SINUS CAFFEINE24 b ANALYSIS SDNN RMSSD SDSD NN5O PNN5O LF LEN HF gt 39 IDKEYWORD MASON I Sooo O OG OC CO C Quo IDKEYWORD WORD t DESCRIPTION Figure 5 4 Database Schematic Diagram 44 CHAPTER 5 Architecture contain any information on how to do anything it just has a link to the control interface that is used as communication mechanism to access its functionality It is on the root package of each visualization resource Visualization Interface This object will contain the functions that the Visualization will require to implement This interface is used by the control agent to interact with the visualization It is under the package imp of each visualization Visualization implementation This class is responsible for imple menting all the functions defined in the Visualization Interface It will also initiali
37. es the user form installing third party software and for its easy encryption properties Database normalized model is shown in Figure 5 4 As we explained in last section ECGData and ECGInfo are stored as objects in a BLOB Binary Large Object field to accelerate queries and updates as well as to reduce database size Our first approach was to create a separate table with ECGData and its values but after some runs we observed t hat the size was increasing too fast In fact it was more than a hundred times bigger than it is now Same happens with RRDetection The result is a simple model with three tables that is efficient in disk usage Even though compression could be made in the serialized object in order to reduce more object space We left this feature to be developed in future versions 5 5 Visual Component Architecture far we have explained what functionalities we require and what struc ture the utilized information has but we still need to implement Use Cases from their abstract definition to well defined components we mentioned in the begging of the chapter the MVC paradigm has guided our steps for the implementation The visual part is separated from the control so that it could be easily modified improved or replaced without the need of rewriting the controller or vice versa Visual Components follow the structure illustrated in Figure 5 5 Each visualization will have different implementations based on its n
38. etection When it is done a new window appears with a chart containing ECG data User presses the mouse left button far enough from any interval extreme inside the chart RR interval is modified where the user releases the mouse If the mouse is released far enough from previous and next intervals a new RR interval is created 1 Special requirements Being Validated User a Preconditions Having an ECG with the RR interval detection done Postconditions Modificacions on the RR Interval detection will be stored to the data model T his will not permanently save information clicking the save RR option will be required 5 3 10 Save RR detection Brief Description This feature brings the possibility of saving the RR detection after any modification made a Event flow enumeration 1 2 3 User request to save RR interval detection RR detection is saved in the database information message is displayed to the user saying that it was successfully saved a Alternate event flow enumeration T 2 User request to save RR interval detection RR detection fails to be saved in the database 36 CHAPTER 5 Architecture 3 error message is displayed to the user saying what happened Special requirements Being Validated User Preconditions Having an ECG with the RR interval detection done Postconditions Modificacions on the RR Interval detection will be stored to the database per
39. event flow enumeration 1 Validated user requests compression 2 Database fails to compress compressed 3 Error message is displayed with failure reason Special requirements Being Validated User Preconditions None Postconditions if successful database size is reduced to its minimum 5 4 Data Model Every application uses data storage in order to work Some static pro grams rely in volatile memory because they do not require permanent stor age but this is not our case ECG2HRV needs to record ECG information as well as its signal and other data in the disk so that they can be queried and accessed later In first place we will describe the structure that ECG2HRV handles this is its class domain In Figure X and Figure Y all the classes and its relations are represented using a class Diagram The following subsection will explain the use and importance of each illustrated object 5 4 1 Class Description ChangePassword Stores the value of the old and new password in an array of characters so that the contained information can be safe ly removed after use Be aware that strings are deleted using JAVA garbage collector and may remain in memory until they are no longer needed and other object overwrites its content 5 4 Data Model 39 Serializable Serializable kasa ST ECGData duration List lt Double gt from List lt Integer gt
40. g is in the filter the full database will be displayed New Search Selecting this action will reset all the search parameters to its default value and erase previous results Modify a search has been performed and an ECG has been selected in the table this will show a frame that will let the user modify any of the chosen ECG properties Open ECG As with modify if a search has been performed and an ECG has been selected in the table a new window will appear display ing the stored signal and the NN intervals In case the NN intervals were not detected before an automatic process will do it Delete This will delete the currently selected ECG on the table from the database HRV Report This click will show detailed information of the HRV parameters for the selected ECG in the table If no NN detection was made for this record RR detection will be performed and shown to the user a Exit This action will close Search Frame and show again the main menu HRV Report This option will allow the user to export the results of the search along with all the ECG properties except for the ECG data to a CSV file The middle table can be reordered by clicking on any of the headers It also allows selection of any of the visible rows so that operations like modify or HRV report generation can be done on that chosen record The tabbed panel at the bottom allows the user to filter the search by changing any of the fie
41. high cut off frequency Muscle artifact is removed at high frequencies but random generated tremors that lay in the frequency of 2 20 Hz can not be separated from the original signal Not all artifact can be removed ECG segments can be targeted by un expected patient movements lead misplacements or muscle tremors Even though the algorithm described later will not count RR intervals that are not considered as normals which solves this issue The filter was implemented in JAVA using an Infinite Impulse Response IIR algorithm shown in formula 3 1 where P is the filter order b are the numerator filter coefficients a are the denominator filter coefficients 2 is the input signal and y n is the output signal The coefficients for a and 5 where calculated using Matlab and stored staticaly in te source code remove the delay from the filter the output of the bandpass IIR But terworth filter was reversed then filtered again with the same parameters and reversed again Figure 3 1 a shows the filtered ECG signal Note that the AC artifact that made the signal go up and down at high frequencies is removed as well as baseline wonder Now we can fully apreciate all the features of the ECG P wave at the beggining RS complex in the middle and T wave at the 3 3 Preprocessing Filtered Signal 11 end of each beat yin y bir n il gt bjx n l 3 1 0 lt i lt P 1 lt j lt P 3 3 Preprocessing Filtere
42. his node delays the electrical excitation to ensure that the atria fully contract and fully charge the ventricles with blood Subsequently the electrical signal travels down two major bundle branches through the center wall of the heart called the ventricular septum These bundle branches are insulated so that the elec trical signal does not excite local septum myocardium As the fibers reach the tip called the apex of the heart they exit the insulating sleeves and electrically excite the ventricular muscle mass 6 CHAPTER 2 Background Upper chambers are the first to contract Lower chambers contract second SA node Larger ventricular muscle ne mass causes larger electrical waveform Smaller atrial muscle mass causes smaller electrical waveform QRS complex IR T wave Lower chambers contract from the bottom and squeeze upward The Purkinje fibers are insulated until they reach the tip of the heart At this point they exit the insulation and excite the muscle mass of the ventricles His Purkinje system Figure 2 1 The graphic on the upper left shows the outer structure of the heart for anatomical reference with part of the heart wall cut away to show the location of the electrical fibers The graphic on the lower right is a schematic of the activation sequence for these fibers and their electrical manifestation on the body surface Excitation wave starts at the apex and moves up towards the upper part of
43. ill but will not be required Additionally key words will b available to create and associate in order to add any other non static field Event flow enumeration 1 Visual input is offered to the user 2 Validated User enters the desired associations with the only re quirement of correctly imported ECG data 32 CHAPTER 5 Architecture 3 Validated User decides to save and detect RR intervals 4 Data is saved Visualization hides and RR detection is started Alternate event flow enumeration 1 Visual input is offered to the user 2 Validated User enters the desired associations with the only re quirement of correctly imported ECG data 3 Validated User decides to save and enter a new ECG 4 Data is saved and state transitions to the first Special requirements Visual input methods will be offered to sim plify this part of the process Preconditions Being Validated User connected to the database and request of the user to insert ECG Postconditions If successful saves the ECG and offers the next se lected option 5 3 5 Modify ECG Brief Description After an ECG has been inserted the data is stored in the database T his feature will enable the user to modify that infor mation using a visual interface Event flow enumeration 1 Validated User selects an ECG and requests its modification 2 new visualization window appears like the one that inserts ECGs but with the values
44. ional state like happiness to other like sadness have been successfully detected and predicted by correlating various HRV markers Wu et al 2010 Despite all the applications and studies in which HRV has utilized its usefulness as research tool remains unexploited in many fields in which rela tions between the investigation and HRV parameter have not been explored We noticed that there was no software allowing the possibility of storing all ECG recordings in order to make sense of extracted HRV data and so we have created a tool to encourage researchers to discover new lines of inquiry In this report we present the development of a new clinical tool ECG2HRV based on personal computing devices T his software allows the user to au tomatically analyze an electrocardiograph ECG for heart rate variability HRV and store it in a database along with all the relevant non demographic patient information For this purpose we developed a new QRS detection al gorithm with low complexity and implemented it in order to give the user a time and frequency domain HRV analysis transformation algorithm to resample at 4 Hz proposed by Berger et al 1986 is applied to the time irregular signal in order to convert the input to a time continuous wave that can be now subject of Fourier analysis in the frequency domain This process is transparent to the user showing only the visual RR detection and HRV report saving time in this way to the researche
45. ith applications s data model and last part will describe ECG2HRV component architecture and the relations that each module has 5 2 Definitions 5 2 1 Use Case Fragment of functionality of the system that enables the user to perform an important action in the application They are detailed in the Use Case Document 27 28 CHAPTER 5 Architecture 5 2 2 Class Domain Description of the structure of ojects that the program uses to exchange and store information as well as their connections associations and com positions It is represented by a Diagram class in UML 2 5 5 2 3 Architecture Essential organization of a system expressed in its components the re lations between them in the application environment and the principles that guide his design and evolution 5 2 4 Component Diagram Visual depiction of the relations that a component has with the rest of the architecture using UML 2 5 standards 5 2 5 Use Case Diagram Visual depiction of the segmented functionalities of an application fol lowing UML 2 5 standards 5 2 6 Model View Controller MVC refers to the software architecture pattern that divides the visual ization from the control agents and the data model used within the applica tion This structure separates each component in different parts so that can be reused and easily modified to correct bugs or to integrate new features 5 3 Use Case Document This section will explain the functiona
46. lds Every field is deactivated by default so that the search has no filters To enable a particular field just click on it and set its value If it is an input field like age click on the button on its left to activate or deactivate it To set HRV analysis parameter filtering click on the HRV Analysis parameter tab and set the desired values All changed properties will be included for the search This means that all fields that haven t got its default value will be concatenated in the search with conjunctive condition AND except for the keywords which follow disjunction OR logic 4 5 RR Detection 23 4 5 RR Detection When inserting an or after a search the user can select the option to view RR intervals window like the one shown at Figure 4 4 will appear window is composed by a top option panel and a Graph displaying selected ECG data and the obtained RR intervals The upper panel allows the following options Previous Misdetection Clicking this action will navigate through the graph to display the previous miss detection or ectopic beat Next Misdetection Same as the option above but showing the next one Save Changes Use this button to store in the database the changes made to the NN detection z Reset RR detection This procedure runs the default RR detection al gorithm disregarding any performed modifications The data will not saved until Save button is clicked HRV report Shows the
47. lity of the application dividing each part in Use Cases The general overview is shown in Figure 5 1 There are two possible actors a Validated user this is the one that has already logged in and the rest normal user beccomes validated when successfully inputs the proper identfication In general the application Use Cases allow insertion modification search deletion of the data the applicaiton stores The second functionality is to permit automatic HRV analysis which needs NN Interval detection We al so though that visualization could be useful to gain the trust of the users by showinf them what is being done and included it as requirement The following subsections will present each Use Case with more detail Each part will contain a brief description to explain its functionality and 5 3 Use Case Document 29 importance event flow ennumeration special requirements preconditions and postconditions Figure 5 1 Global Use Case diagram 5 3 1 Login Brief Description this Use Case encapsulates the functionality re quired to start the application and make sure that the user is the owner of the data contained in the persistence by validating with a password dialog or other methods IT is also responsible for starting the required components that will make the application work after its identity success 1 Event flow ennumeration 1 The application starts and shows an input dialog 2 The application receives
48. manently 5 3 11 HRV Analysis z Brief Description This is the second most important Use Case of the application It is responsible for converting the NN intervals from the RR detection to HRV parameters in time and frequency domain using the techniques explained in Chaptter 3 Event flow enumeration 1 User request HRV analysis for an already RR Interval detected ECG 2 Program runs required algorithms to obtain time and frequency parameters 3 Results are displayed with charts and text inside a new a frame Alternate event flow enumeration 1 User request HRV analysis for an ECG without RR detection made 2 RR detection Use Case is executed for that algorithm Special requirements Being Validated User Preconditions Having an ECG from the database selected a Postconditions HRV is done if RR detection is too 5 3 12 Save HRV Analysis Brief Description This feature allows to store calculated HRV pa rameters in the database a Event flow enumeration 1 HRV analysis is performed on an ECG 2 data is saved in the database Alternate event flow enumeration 5 3 Use Case Document 37 1 HRV analysis is performed on an ECG 2 data fails to save in the database 3 An error message is shown with the reason of failure Special requirements Being Validated User a Preconditions Having an HRV analysis performed on an ECG stored in the database a Postconditions if successful HRV analy
49. nd Drug Administration and can be obtained at a quite reasonable prize this is less than 200 This new 1 4 History example of a potential ECG2HRV user 3 hardware pieces have simplified clinical information gathering by enabling research to move through different paths easily without high budget require ments Taking safe ECG measurements from patients now becomes as easy as to let them hold a phone with a piece of hardware for five minutes Not only researchers are the only sector designated for the use of ECG2HRV but anyone with access to any of these devices curious enough about HRV and their conditions is also a perfect target They could even conduct re search taking ECG of themselves at different situations and scenarios 1 4 History example of a potential ECG2HRV user Paul is a fifty three years old cardiologist that worked for twenty years diagnosing diverse abnormalities related to heart diseases He has always dreamed on conducting research HRV but never had enough evidence of where to start in After twenty years of work he has realized that there may be a link between Vagus nerve stimulation and improvement in heart rate variability He came to that hypothesis after taking five minutes ECG recordings from patients that used listening aid devices His guess is that a small intensity of the electric field generated from this hardware is conduct ed through Tragus and absorbed by the parasympathetic Vagus nerve In his
50. nd user manual describing how to use the ap plication If the reader is not a developper he will probably only be interested in this section Chapter 5 presents the architecture of the application following the standard UML Unified Modeling Language notation This part is conceived for other developpers to diminish the burden of modifying it to include new features Chapter 6 Will disscuss advantages and shortcomings of ECG2HRV Chapter 2 Background One s mind has a way of making itself up in the background and it suddenly becomes clear what one means to do A C Benson ABSTRACT This chapter will explain the required concepts to un derstand what concepts and tools are used by the application as well as the relevant physiological information that is attached to them 2 1 What is an ECG From the body s perspective the heart s function is that of a hydraulic pump moving blood through the vessels that course through the body In order to accomplish this the heart has both mechanical and electrical struc tures that have important characteristics related to health and illness heart consists of four chambers two upper chambers left and right atriums and two lower chambers right and left ventricles From a mechan ical standpoint the muscle mass of the two atria contracts when the SA Node first excites the local myocardial tissue There is a delay point between the upper and lower chambers called the AV Node T
51. nitializes those components and gives the interfaces they require It needs a Validated User so that database is connected and can perform any search in the rest of the components z Use Cases implemented InsertECC Just the request part SearchE CC Just the request part Utilities Just the request part Exit Validated User Ae WO N LR 5 6 Control Component Architecture 47 Main Me O IttvisMenuPrincipal Required Initialized components 5 Be eo Menu principal MARE id ModDB 9 IffModD8 ItfConMenuPrincipal ____iconMenuPrincipsi Sp Figure 5 7 Component Diagram of Main Menu control 5 6 3 Insert ECG 1 Brief Description This component is responsible for giving the user the option to InsertECG or ModifyECG successful login The compo nent requires Search ECG in order to go back to the same search upon modification completion z Use Cases implemented 1 InsertECC 2 Modify ECG 3 DetectRR just the request part 48 CHAPTER 5 Architecture emp Inserteca 7 ItfModDB Required Initialized components ItfConinserterECG 2
52. ny of its func tions in order to be able to fit in this page 5 4 Data Model 41 1 ECG This class represents the core of class domain It stores the value of a single ECG record It contains an ECGlnfo object as well as a list of keywords representing the associations of that particular ECG to clinical or other features T his object also stores the value of the raw ECG RR detection and HRVAnalysis If any of those have not been made yet its value will be null 1 ECGData This container will store the raw ECG data It contains two array of double with the time and the values for each sampling point It is serializable this means than can be easily saved in JAVA object format T his was implemented in order to store the full ECG in the database in a single record instead of thousands By doing that we prevent database size from growing exponentially and permit accessing data faster too Time is not saved in the database in order to occupie less space Instead it is recreated from the sampling frequency starting by zero a ECGInfo Contains all relevant medical characteristics detailed in the Userguide section4 3 ECGSearch This class was created so that all the search filtering vari ables where included in a single class It contains an ECGInfo variable for the value of default properties two HRVAnalysis items to store the value of gt and lt parameters list of keywords with no length re striction is also include
53. om other cultures and their methods It was not until this year when I studied in the University of Oklahoma that my brain developed a research way of thinking Without that cultural exchange I would probably had not even thought about building something like this project or other I am working on but thanks to that European scholarship I was able to adapt to a different University system and blend it with the Spanish one in order to get best of both I hope that this kind of opportunities will never disappear from the edu cational budget of Spain or any other country If they are given to students with thirst of knowledge they will make those seeds grow along with the prosperity of the nation finish I would like to credit my directors Dr John and Dr Hidalgo for their guidance an continuous help in this project Education is the most powerful weapon which yow can use to change the world V VI SPECIAL REGARDS Nelson Mandela Abstract Man is a tool using animal Without tools he is nothing with tools he is all Thomas Carlyle Heart Rate Variability HRV is a non invasive and simple source of markers that have been proved to be clinically relevant in studies to predict patient evolution like survival after Myocardial Infarction Buccelletti et al 2009 and classification of others like Diabetic Neuropathy Pagani 2000 Further more Medicine is not the only field it has being applied Transition from one emot
54. on stored in database and performs it 3 RR detection is saved to the database and displayed andmodifi cation of RR inervals is called Alternate event flow enumeration 1 Validated User requests RR interval detection for a particular ECG from Search or new Insert 2 The program checks that ECG has already an RR Interval detec tion stored in database 3 RR detection is retrieved and odification of RR inervals is called Special requirements Being Validated User Preconditions Being inserting or modifying an ECG record ECG data and Sampling rate is known a Postconditions RR detection is made and stored in database if need ed ew window is then shown allowing its modification 5 3 9 Modify RR detection Brief Description This option will offer the required visual tools to modify the detections obtained to correct possible miss detections Event flow enumeration 5 3 Use Case Document 35 User request RR interval detection When it is done a new window appears with a chart containing ECG data User presses the mouse left button near the beginning or the end of an interval inside the chart RR interval is modified where the user releases the mouse If the mouse is further away than previous or next RR interval the R peak disappears and the new RR interval is considered to be from past RR interval to the next a Alternate event flow enumeration User request RR interval d
55. r and enabling storage of the results with all the ECG relevant information so that it can be used to cross search groups of patients with certain clinic features VII VIII ABSTRACT Key Words HRV ECG ECG2HRV HEART RATE VARIABILITY DATABASE TIME FREQUENCY DOMAIN Resumen La Variabilidad de la frecuencia card aca VFC es una fuente de mar cadores que ha demostrado una clara relevancia cl nica en la predici n del Infarto de Miocardio Buccelletti et al 2009 y en la clasificaci n de otras enfermedades tales como el trastorno neurop tico Pagani 2000 Es mas la VFC no s lo se emplea nicamente en el campo de la Medicina Diversos marcadores de la VFC han sido correlacionados satisfactoriamente con di versas transiciones emocionales tales como el paso de la alegr a a la tristeza Wu et al 2010 Independientemente de los estudios en los que la VFC ha sido empleada su potencial como herramienta de investigaci n est aun por desarrollar y en muchos campos las correlaciones con la VFC permanecen inexploradas Al darnos cuenta de que no se dispon a de software adecuado que permitiera el almacenamiento de registros de Electrocardiogramas ECG para poder extraer a partir de ellos datos tiles de VFC creamos una herramienta que permitiera impulsar a los investigadores hacia nuevas l neas de desarrollo En este trabajo presentamos una nueva herramienta cl nica ECG2HRV capaz de ejecutarse en cualquier or
56. rtant feature of the application It enables multi field cross related search so that any stored characteristic can be subject to filter the results obtained This enables Validated User to find potential groups with common markers that may lead to new lines of inquiry HRV analysis results and ECG associated information will be responsible for the filtering If the parameter is numeric or it can be ordered in time or quantity range search with gt and or will be available a Event flow enumeration 1 Validated User inputs the desired filtering parameters and re quests the search 2 Database performs the search and returns the results 3 Results are visually displayed and record selection is allowed Alternate event flow enumeration 1 Validated User inputs the desired filtering parameters and re quests the search 2 Database performs the search but has an error 3 An error dialog is shown to Validated User with a plausible reason Special requirements None 1 Preconditions Being Validated User connected to the database and have a visual interface open to allowing search filtering a Postconditions If successful results for the search will be displayed and the following new options will be available modify ECG delete ECG RR detect HRV analysis 5 3 4 Insert ECG 1 Brief Description ECG insertion will permit the user to associate an ECG to the desired characteristics Default properties will be offered to f
57. se Select if the patient suffers from this disease and pick its degree between Mild Modere or Severe If it does not have MVP or it is not know leave selected NONE Smoker Check if the patient is a smoker Caffeine Check if the patient has ingested caffeine in the past 24 hours Alcohol Check if the patient has ingested alcohol in the past 12 hours Sport Check if the patient has practised sport in the past 24 hours a Sinus Check if the ECG is mostly following Sinus Rhytm Cardiomyopathy Check if the patient suffers from a known cardiomy opathy Hypertension Check if the patient suffers from uncontrolled hyperte sion Vaso Vagal Check if the patient suffers from recurrent vaso vagal syn copal episodes that made him faint Depression Check if the patient suffers a depression that is not being currently under treatment Diabetic neuropathy Check if the patient has received a diagnosis of a diabetic neuropathy Myocardial infarction Check if the patient has experienced a myocar dial infarction in the last year Keywords This part will allow you to associate any other feature that is not in the ones introduced above so that the researcher does not loose any relevant information he may want to attach to each inserted ECG If the word is not on the database click on the green cross and a new dialog will allow you to add the desired keywords with a small description Then you will be able to
58. sis is associated to the se lected ECG permanently 5 3 13 Change Password a Brief Description This feature permits to change the current pass word of encryption and access for the database Event flow enumeration 1 Validated user inputs correct old password and a new alphanu meric chain bigger than eight characters 2 Database password is changed to the new password Alternate event flow enumeration I 1 Validated user inputs incorrect old password and a new alphanu meric chain bigger than eight characters 2 Database password is not changed and a message is shown Alternate event flow enumeration II 1 Validated user inputs correct old password and a new alphanu meric chain bigger than eight characters 2 Database password fails to change 3 An error message is shown with the reason of failure Special requirements Being Validated User Having enough mem ory to remember a new password Preconditions Know the new password Postconditions if successful password is changed 38 CHAPTER 5 Architecture 5 3 14 Compact Database Brief Description This feature allows to compress the database in order to safe disk space Persistence may span more size than required after severe changes like modification of ECG data or deletions a Event flow enumeration 1 Validated user requests compression 2 Database is compressed 3 Information dialog is displayed with a message of success Alternate
59. stored in the database 3 Validated User changes the desired information and saves 4 Data is updated and the selected action back to search or view RR intervals is requested a Alternate event flow enumeration 1 Validated User selects an ECG and requests its modification 2 new visualization window appears like the one that inserts ECGs but with the values stored in the database 3 User decides to cancel the modifications 4 Data is not saved visualization is closed and the program goes back to the previous window Special requirements Being Validated User 5 3 Use Case Document 33 Preconditions uniquely identified ECG that still exists in the Database and the request of the user to modify it Postconditions If successful update the database content with the input 5 3 6 Delete ECG 1 Brief Description This feature will enable the permanent removal of any inserted ECG from the database Event flow enumeration 1 Validated User selects an ECG and requests its deletion 2 Its content gets successfully deleted from the persistence Alternate event flow enumeration 1 Validated User selects an ECG and requests its deletion 2 An error message is shown to the user and application continues its run ECG reamins recorded in dataabse Special requirements Being Validated User Preconditions A uniquely identified ECG that still exists in the Database and the request of
60. t is not enough to create a simple deci sion maker in some case scenarios like abnormal QRS due to heart diseases or very noisy segments These situations prompted us to develop a NN Interval detection algorithm The approach is to find the first biggest valley within time threshold THRESHOLDI then detect if there exists a peak within a second time threshold THRESHOLD2 that meets the requirements these are being a peak with one of the highest heights in the differentiated waveform and having a peak with an acceptable height in the original waveform To prevent bigger T waves to be detected all data from last positive QRS detection to 200 ms later is not used given that it is physiologically impossible to have another QRS in such small time lapse classify the different type of RR intervals we chose to implement the same approach used in 6 We have a NN Average calculated with the last eight valid and normal beats and another RR Average of the last eight beats beat is considered to be normal if its duration exceeds 0 92 NN Average but do not surpass 1 16 NN Average All intervals that lay out of normality are disregarded If last eight RR detections are normal then NN Average becomes RR Average 2 last threshold THRESHOLDS was used for the differentiated input to prevent small valleys originated from noise or other sources from being detected as possible beats This value is initialized with the first detections and is up
61. the ventricles In this fashion the ventricles squeeze the blood out into the body The key to ensure that the atria and ventricles operate in the proper sequence and with appropriate timing is the electrical system of fibers in the heart But it is the mass of contractile muscle tissue that produces the electrical signal that can be recorded from the body surface The body surface recording of the cardiac electrical signal is the electrocardiogram often referred to as the ECG 2 2 What is an RR interval The time lapsed from one QRS complex to the next is called RR interval The fiducial mark we have used to determine the QRS complex is the end of the rising edge this is the peak due to its easily recognizable pattern This duration is measured in miliseconds and can be used to calculate the heart rate frequency An example is shown below in 2 2 An special case of RR interval is the normal to normal Interval or NN Interval Synus RR intervals whose neighbours follow synus rhytm too are considered to be Normal Only normal intervals are taken into account to calculate the Heart Rate Variability of the pacient the remaining beats are discarded 2 3 What is HRV Interval R Qs Figure 2 2 Example of RR interval 2 3 What 15 HRV The change in duration between NN intervals due to physiological factors is called Heart Rate Variability In 1996 the European Society of Cardiolo ey along with the North American
62. urn to the last menu without saving any changes made to the information 4 4 ECG Search Once the user has logged in he will be able to search any of the inserted ECGS by selecting Search ECG option in the main menu then window like shown in Figure Figure 4 3 will be displayed S Search ECG 2 5 NN NEW SEARCH HRV REPORT 25 s Gender LF HF VLF ECG Information HRV Analisis parameters Gender X Current Conditions KEY WORDS H ge va H Eismoker currently Exercise over past 24h ja cd la O Caffeine on past 24h Alcohol on past 12h Pactient History Pactient History Heart failure NYHA ORecurrent vaso vagal syncopal episodes Mitral valve prolapse O Major untreated depression or psychosis O Sinus Rhythm O Diabetic neuropathy Gco myocardial infarction Uncontrolled hypertension Figure 4 3 Search ECG Frame 22 CHAPTER 4 2 User Guide In the frame we can observe a top panel where all the possible actions are displayed a white table where search results will be displayed and a tabbed panel with filtering options for the search The first one allows the following actions Search Perform the search on the database using as filter the currently selected options If nothin
63. window displays the results obtained from the process described in Chapter 3 The power spectrum density is shown on the top left chart while the LF to HF ratio is on the bottom left one Time domain results and frequency domain parameters are displayed on the right side of the window 4 7 Utilities Some database options are available through the Utilities option of the main menu window like the one illustrated in Figure 4 6 will be displayed Two options are available Change passwod and Compact database 4 7 1 Change password This option permits the user to set a password of his choice We strongly recommend to do this after the installation or wherever the integrity of such key is not ensured window like displayed in Figure 4 7 will appear in the screen 4 7 2 Compact database After several modifications or deletions database size may increase con siderably This option will compress the information to reduce the size that is occupied by that information It is recommended to do it from time to In the next 25 HRV REPORT HRV Analysis M ECG2HRY E to Heart Rate Variability 0 20 0 25 Voltage mV LF to HF Ratio VLF 103 72 52 LF 115 77 s2 HF LF NLF 28 019 114 95 62 27 821 LF HF 96 50 178 96 LF 50 178 96 e HF 49 822 Figure 4 5 HRV Analysis Frame time or after big database changes It does not take a long time to execute an
64. with a desired ECG time segment Without this skill testing any hypothesis using HRV markers based on obtained data becomes a tedious task that obliterates fu ture research by shadowing data we already have not allowing the physician to easily correlate that information In this reality it was our main purpose to prevent looses of that precious content as well as promoting HRV future research second reason that led us to start this task was the fact that many health professionals do not have the required knowledge time or experience 1 2 CHAPTER 1 Introduction to extract those values from a raw ECG signal This application enables any non related signal processing user to import a single lead ECG with any valid sampling frequency as input and get it filtered processed and analyzed by following three simple steps To ensure user confidence this procedure is backed up with visual feedback so that he can guess what is being worked out in the background 1 2 What is ECG2HRV work presented in this report implements industry standard mea surements of HRV within the construct of a platform independent utility coded in JAVA that allows the clinician to immediately assess a patient HRV status and store this information along with other relevant clinical features HRV parameters are computed in both the time domain and the spectral domain based on a five minute single channel ECG recording The values are presented to
65. y ta time VH 1 1 a ee time Figure 3 2 Figure extracted from Berger et al 1986 b The heart rate samples corresponding to the ECG signal in a determined using our algorithm The num ber of RR intervals within the local window centered at t is a I2 and at t2 is b I3 C 14 The value of the heart rate at each sample point is taken to be the number of intervals that fell within the local window centered at that point divided by the width of the window as described in the text c The corresponding instantaneous heart rate signal The value held during each interval is the reciprocal of the dura tion of that interval The sample values in b are equivalent to those of the signal that would result from convolution of the signal in c with a rectangular window that is two sample intervals wide Variable Units Description Frequency range total spectrum ms Power in all spectrum lt 0 4 Hz power VLF ms Power in very low frequency range lt 0 04 Hz LF ms Power in low frequency range 0 04 0 15 Hz LF norm n u LF power in normalised units LF Total Power VLF x 100 HF ms Power in low frequency range 0 15 0 4 Hz HF norm n u HF power in normalised units HF Total Power VLF x 100 HF LF Ratio ms2 HF ms2 Table 3 2 Selected frequency domain measures for HRV from Camm et al 1996 In the next 15 In the next chapter In this sections we have explaine
66. ze the required resources Resources Final implementation for the visualizations In this cate gory lay all designed frames panels dialogs or any other visual com ponent that the visualization requires It is under the package imp gui of each visualization We will not describe each of the implemented visual components Our reason for this behavior is that visualization is subject to current component library functionality and developer Dependant Instead we will focus on explaining control agents and their interactions and requirements which are less likely to change 5 6 Control Component Architecture 45 Figure 5 5 Visualization component model 5 6 Control Component Architecture The Use Cases described in this chapter at section 5 3 have been imple mented with the following components 5 6 1 Login Brief Description This component is responsible for initializing database Component interface and Main Menu upon successful login z Use Cases implemented 1 Login 2 Exit User 46 CHAPTER 5 Architecture Required Initialized components ItfConMenuPrincipal Created Components Figure 5 6 Component Diagram of Login control 5 6 2 Main Menu Brief Description This component is responsible for giving the us er the option to start any of the three following options InsertECG SearcECG or Utilities It also i

ECG2HRV: Electrocardiagram to Heart Rate Variability - E

Contents

Download Pdf Manuals

Related Search

Related Contents