Detection of Cardiac Murmurs

The project describes a system for detecting cardiac murmurs so people are aware about their heart condition. The murmurs are the pathologic heart sounds that are produced because of turbulent blood flow that is enough to cause audible noise. Murmurs can be detected by a stethoscope but the accuracy level is not satisfactory. Different methods have been developed for detection and all are efficient enough but looking cost aspect a system should be designed that detect murmurs easily with maximum accuracy level. The designed system calculates the low energy rate (LER) from the recorded cardiac signals and then classifies the signals and either normal or murmur signal. The system is completely compatible with personal computer or a laptop for data visualization. Under the experiment a large number of people have been tested and results are consulted with a physician too. Results described for system designed to detect murmur classify signal either normal (LER >0.8) or murmur (LER <0.8) Background Many studies have worked towards designing system for many detection of cardiac murmur and improving systems accuracy to have better results. Cardiac murmurs can mostly be detected with the help of a simple stethoscope but the results are not satisfactory and the difference process requires special equipments and specialized trained physician. So there is a requirement of a system that can effectively detect cardiac murmurs and that is easy to be implemented. A computer aided diagnosis algorithm has been designed and implemented on an ultra cyclone II FPGA to detect cardiac murmurs from recorded heart signals, but system is difficult to implement because of use of FPGA board and also it is difficult to interface with computer system. This intent to design a system that is easy to implement, simple to construct. Thus, this research is to develop an electronic stethoscope, that directly fed the cardiac signal to the sound card, amplify the sound and classifies them either normal cardiac signal or cardiac murmurs on the basis of analyzed signal parameters. Thus, important goal is to achieve better accuracy and to design cost effective and simple system. Introduction Murmurs The murmurs are the pathologic heart sounds that are produced because of turbulent blood flow that is enough to cause audible noise. Anju et al, International Journal of Computer Science and Mobile Computing, Vol.3 Issue.7, July2014, pg. 81-87 © 2014, IJCSMC All Rights Reserved 82 Mostly murmurs are normal and can be heard with the help of a stethoscope, during a physical exam and do not require any treatment. Cardiac murmurs, sometimes called as pathologic murmur as they are a result of some problems like narrowing of valves, leaking of valves or presence of abnormal passages from which blood flows in or near the heart [ E. Etchell et al, 1997]. However, functional murmurs are heart murmurs due to physiologic conditions outside the heart, so they are also termed as physiologic murmur. Physiologic murmur also referred as innocent murmur as they are occurring under normal conditions causing no problem. But, pathologic murmurs, should be evaluated by an expert. Heart murmurs can also be caused if blood is flowing through any damaged or over worked heart valve [A.R. Freman et al and D. Mason]. The valvular problems may be by birth or may occur later because of ageing process or other related problems as heart attacks, rheumatic fever. Heart murmur can also be classified by using following parameters-frequency, pitch, quality, characteristics, location, shape and duration of murmur. Normal heart signal consists of two sound signal that corresponds to lub and dub phase and these are termed as S1 and S2. There are some other signal activities between S1 and S2 and these are the abnormal sound signals. If we are able to locate the S1 and S2, then individual heart signals can be identified. Further on the basis of timing cardiac murmur can be –systolic and diastolic. Intensity can be graded on 6 point scale, where intensity increases from grade 1 to grade 6 type of murmurs. Different type of heart valve diseases are – Mitral valve prolapse, in which mitral valve closes completely when left ventricle of the heart contracts, preventing blood from flowing back to left atrium. If any part of valve bulges out so that it does not close properly the mitral valve prolapse. This situation is not so serious but it can result into regurgitation (backward flow of blood). Mitral or Aortic Regurgitation – Backword blood flow occurs with mitral valve prolapse or mitral valve or aortic stenosis. For counter acting this back flow of blood, heart have to work harder to force the blood. But with passage of time, this weakens the heart and leads to heart fai lure. Mitral valve or Aortic stenosis These occur as calcium get deposited on valves with age. Mitral or aortic valve both on the left side of heart, become narrowed because of infection (rheumatic fever), or may be narrowed by birth. In this condition heart to work harder to pump enough blood to satisfy the oxygen needs, this leads to heart failure. Aortic sclerosis –It is due to thickening, scarring or stiffening of the aortic valves. This situation is not so serious as the valve can function for years after it is detected. Congenital heart defects Every year thousands of cases arises having heart defects by birth, these are congenital heart defects. Heart murmur can also be heard by using a stethoscope, then doctor prescribed further tests to check whether it is innocent or by congenital defect. Following tests are ordered by the doctor to see whether heart murmur is innocent or caused by acquired valve disease or congenital defect. ECG (Electrocardiogram)Which measures the electrical activity of the heart. Chest X-Rays to see if the heart is enlarged due to heart or valve disease. Echocardiography which uses sound waves to map the heart structure. Most murmurs can be heard with the help of stethoscope but the results with a normal stethoscope are doubtful. A system based on segmentation techniques and artificial neural network has been implemented as a detector and classifier. The system allows the user to design and create a heart sound classifier to enable the user to select any audio file to be used as input to ANN system [S. L. Strunic]. To show systems ability testing performed to classify type of heart sounds-normal aortic stenosis and aortic regurgitation. An automated cardiac auscultation system implemented for detection of pathologic heart murmurs. The cardiac sound signals were low pass filtered at 1000 Hz and converted to time scale domain using continuous wavelet transform method. The derived energy values, expressed in decibels are relative. Data were analyzed using one – tailed, two sample unequal variance student’s t-test, with p< 0.005 [W. R. Thompson et al, 1993]. A computer-aided diagnosis (CAD) algorithm has been designed and implemented on an Altera Cyclone II FPGA to detect cardiac murmurs from recorded heart signals.The FPGA system interfaces with a commercial digital stethoscope to acquire real time data as well as a VGA(video graphics array)compatible monitor for visualization and metric reporting[Michael Yenting,2010 ]. Signal processing approach has also been used to isolate systolic heart murmurs based on wavelet transform and an energy index[B. Tovar-Corona et al,1999].This approach demonstrates the isolation of the systole interval and the detection of systolic murmur onset and duration[Nikolay Atanasov and Taikang Ning,2008].one of the methods uses artificial neural network , the method include the major steps as study subjects, recording of the signal, signal analysis and ANN prediction. Using an electronic stethoscope, heart sounds are recorded from 69 patients (37 pathological and 32 innocent murmurs). Sound samples are processed using digital signal analysis and fed into a custom ANN. With optimal settings, sensitivities and specificities of 100% were obtained on the data collected with the ANN classification system developed. in the process to Study Subjects Digital heart sound recordings are obtained from various pediatric patients who were referred to a cardiology clinic for evaluation (aged 1 week to 15 years; mean age, 2 years). in the process of recording signal :using an electronic stethoscope and a personal computer ,heart sounds were recorded at the left of midsternal border in the supine position in a standard examination room. For each patient, 2 separate heart sound recordings were acquired. 2 heart sound recording are equivalent to 8 heart cycles in length. one sound sample of three representative heart cycles is chosen form these two recordings. One recording location was chosen with the expectation that murmurs generally not heard from this location (using standard stethoscopic methods) would be detected by the electronic microphone used due to its sensitivity. during Signal Analysis process in Artificial Neural Network customized ANN software developed in the laboratory. ANNs were trained to discriminate between normal and pathological examples. for Statistical Analysis Standard equations for sensitivity and specificity are used ANN Predictions to differentiate between murmur and normal cardiac murmur. this method had certain limitation because The process of selecting the optimal 3 consecutive heart cycles required an intelligent selection process. In developing a screening device, automating this selection process will need to be investigated. The performance of the ANN with the addition of noise processes (eg. from an uncooperative patient) to the input signal will need to be investigated. Anju et al, International Journal of Computer Science and Mobile Computing, Vol.3 Issue.7, July2014, pg. 81-87 © 2014, IJCSMC All Rights Reserved 83 In order to detect sound, recordings of diastolic heart sound segments are analyzed by using four signal process-ing techniques; the Fast Fourier Transform (FIT), the Autoregressive (AR), the Autoregressive Moving Average (ARMA), and the MinimumNorm (Eigenvector) methods. To further enhance the diastolic heart sounds and reduce background noise, an Adaptive filter is used as a pre-processor [Yasemin M. Akay,2013 ]. The power ratios of the FFT method and the poles of the AR, ARMA, and Eigenvector methods are used to diagnose patients as diseased or normal arteries using a blind protocol without prior knowledge of the actual disease states of the patients to guard against human bias. Results showed that normal and abnormal records are correctly distinguished in 56 of 80 cases using the Fast Fourier Transform (FFT), in 63 of 80 cases using the AR, in 62 of 80 cases using the ARMA method, and in 67 of 80 cases using the Eigenvector method. Among all four methods, the Eigenvector methods showed the best diagnostic performance when compared with the FFT, AR, and ARM A methods. These results confirm that high frequency acoustic energy between 300 and 800 Hz is associated with coronary stenosis . Method The procedural steps include real time signal acquisition and filtering of the signal, display of the signal waveform, classification of the signal as normal heart signal or a cardiac murmur and the performance metrics calculation. HUMAN SUBJECT AORTA ELECTRONIC STETHOSCOPE SOUND CARD MATLAB SOFTWARE SIGNAL ACQUISITION & FILTERING CLASSIFICATION OF SIGNAL AS NORMAL OR CARDIAC MURMUR WAVEFOR M DISPLAY PERFROMANCE METRICS CALCULATION Fig: System design For the signal acquisition purpose a digital stethoscope is used. The stethoscope is a medical device used for listening hear t sounds of human body and for measurement of blood pressure. An electronic stethoscope amplifies body sounds by electronically amplifying the signal. Electronic stethoscope converts sound waves to electric signals which can be amplified and processed. The signals can be interfaced with a computer to analyse the signal waveforms. The signal processing approach is used to perform the cardiac signal analysis and acquisition. An electronic stethoscope is designed to acquire cardiac signals in real time. Signals transferred to the sound port from where they are transferred to the sound card. The MATLAB software used is compatible with sound port and the signals are acquired directly from the sound port to the software. The cardiac signal from the human aorta is acquired in real time using the digital stethoscope. Aorta is the major artery of human body that carry oxygenated blood to all the organs of body else to lungs. The cardiac signals are directly fed to the sound card. No external interfacing circuitry is required for interfacing hence it reduces the system cost. From the sound card, the signals are transferred to MATLAB software. The software has the provision to acquire signals from sound card .after the signal acquisition in the software; the noise is removed that is the process of filtering is applied on the acquired signal. The raw cardiac signal before filtering and the filtered signals are displayed on the software. After analysis of signal parameters the cardiac signal is classified as either normal or a cardiac murmur. The performance metrics such as accuracy, specificity, and sensitivity of the algorithm are calculated in order to be sure about the results of the algorithm. Parameters: RMS: root mean square value. The root of mean of square of all the values of a signal is termed as root mean square value of the signal.