Collected by Laura Malave - http://www.eng.usf.edu/igert/students/lmalave/stuf/lmalave_rsch_pg7.htm
Abdel-Alim O, Hamdy N, and El-Hanjouri MA. Heart Diseases Diagnosis Using Heart Sounds. NN Radio Science Conference, 2002.
In this paper a feed forward artificial neural networks were used to classify several valve-related heart disorders. The heart sounds were recorded via the traditional stethoscope. Relevant features were extracted using several signal processing tools, discrete wavelet transfer, fast fourier transform, and linear prediction coding. Recognition rates achieved were around 95.7%. The raw data, a 1-minute long record of heart beats representing each disease, was processed to extract the necessary features by: a) denoising using wavelet analysis, b) separating one beat out of each record c) identifying each of the FHS and the SHS. It was found that diseases related to valve problems could be classified according to the time separation between the FHS and th SHS relative to cardiac cycle time, namely whether it is greater or smaller than 20% of cardiac cycle time. For the first group the NN comprises 6 nodes at both ends, with one hidden layer containing 10 nodes. For the second group, the resulting 216 LPC coefficients for each event were fed to two separate neural networks containing 50 hidden neurons. The NN for classifying systolic diseases should have 4 output neurons, those for classifying diastolic diseases can contain 2 neurons only. For the first group the training set consisted of 650 cases, testing set consisted of 200 cases and had a recognition rate of 95%. For the second group the training set consisted of 320 casses and 100 for testing.
Douglas A. Balster, BS, David P. Chan, MD, Daniel G. Rowland, MD and Hugh D. Allen, MD. Digital Acoustic Analysis of Precordial Innocent Versus Ventricular Septal Defect Murmurs in Children, The American Journal of Cardiology, Volume 79, Issue 11, 1 June 1997, Pages 1552-1555
Barschdorff, D. et.al. Automatic Phonocardiogram Signal Analysis in Infants based on Wavelet Transforms and Artificial Neural Networks. Computers in Cardiology, 1995. pp. 753-756.
Barschdorff, D.; Ester, S.; Dorsel, T.; Most, E. Neural network based multi sensor heart sound analysis. Computers in Cardiology 1990. Proceedings. , 23-26 Sept. 1990 Pages:303 - 306.
Bhatikar SR, Mahajan RL, DeGroff C. A novel paradigm for telemedicine using the personal bio-monitor. Biomed Scie Instrum 2002; 38: 59-70.
Cathers Ian. Neural Networks Assisted Cardiac Auscultation. Artificial Intelligence in Medicine. 7 (1995) 53-66.
Chiu CC ; Chang HH ; Yang CH. Objective auscultation for traditional chinese medical diagnosis using novel acoustic parameters. Comput Methods Programs Biomed (Computer methods and programs in biomedicine.) 2000 Jun; 62(2): 99-107
Dahl L.B., et.al. Heart murmurs recorded by a sensor based electronic stethoscope and e-mailed for remote assessment. Arch Dis Child 2002; 87:297-301.
DeGroff CG, Bhantikar S. et.al. Artificial Neural Network-Based Method of Screening Heart Murmurs in Children. Circulation. 2001;103: 2711-2716.
The aim of the present study was to train an Artificial Neural Network to distinguish between innocent and pathological murmurs effectively. The heart sounds of 69 patients (37 pathological, and 32 innocent) were recorded with an electronic stethoscope, Cambridge Heart Sound Microphone, at one location (left midsternal border). The optimal 3 consecutive heart sounds were manually specified. For signal analysis, a normalized energy spectrum of the sound data was obtained by applying a Fast Fourier Transform. The various spectral resolutions and frequency ranges were used as inputs into the ANN to optimize these parameters to obtain the most favorable results. A customized artificial neural network was used, this was a feed-forward ANN using the back-propagation learning algorithm. Because of the large data size the jack-knife method was used. With optimal settings sensitivities and specificities of 100% were obtained.
related: Larkin M. Paediatric heart sounds assessed by computer. The Lancet, Volume 357, Issue 9271, 9 June 2001, Page 1856.
El-Hanjouri M, Alkhaldi W, et.al. Heart Diseases Diagnosis Using HMM. IEEE Melecon 2002.
This paper suggests a diagnostic technique for heart diseases using heart sounds. The heart sounds were denoised using six-stage wavelet decomposition, thresholding, and then reconstruction. Three feature extraction techniques were used: 1. traditional method, the Decimation method, and the wavelet method. Classification of the heart diseases, 11 different, is done using Hidden Markov Models (HMMs). The HMM was implemented using the HMM HTK toolkit. Results given the 3 different feature extraction methods were 97.3%, 98.2%, and 99.1%, respectively. This is a better performance than has been achieved through neural networks, 95.7.
Folland R ; Hines EL ; Boilot P ; Morgan D. Classifying coronary dysfunction using neural networks through cardiovascular auscultation. Med Biol Eng Comput (Medical & biological engineering & computing.) 2002 May; 40(3): 339-43.
Grewe K. and Crawford MH. Differentiation of cardiac murmurs by dynamic auscultation. Current Problems in Cardiology, Volume 13, Issue 10, October 1988, Pages 673-721.
Hebden, J.E.; Torry, J.N. Identification of aortic stenosis and mitral regurgitation by heart sound analysis. Computers in Cardiology 1997 , 7-10 Sept. 1997 Pages:109 - 112.
Hadjileontiadis, L.J.; Panas, S.M. Discrimination of heart sounds using higher-order statistics. Engineering in Medicine and Biology society, 1997. Proceedings of the 19th Annual International Conference of the IEEE , Volume: 3 , 30 Oct.-2 Nov. 1997. Pages:1138 - 1141 vol.3.
Haghighi-Mood A, and Torry JN. Time-Frequency Analysis of Systolic Murmurs. IEE, 1997 pp. 71-72.
Hall LT, et.al. Sensor System for Heart Sound Biomonitor. Microelectronics Journal 31 (2000) 583-592.
Hayek, CS. et.al. System and Method for Diagnosing Pathologic Heart Conditions. Patent Application 20030055352. March 2003.
Hayek CS. Thompson WR, et. al. Wavelet processing of systolic murmurs to assist with clinical diagnosis of heart disease.Biomed Instrum Technol (Biomedical instrumentation & technology / Association for the Advancement of Medical Instrumentation.) 2003 Jul-Aug; 37(4): 263-70
Hong Wang; Le Yi Wang. Multi-sensor adaptive heart and lung sound extraction. Sensors, 2003. Proceedings of IEEE , Volume: 2 , 22-24 Oct. 2003
Pages:1096 - 1099 Vol.2.
Hui, W.W.; Pitt, R.A.; Matonick, J.P.; Li, J.K.-J. Comparison of heart sounds recorded at the chest and a remote arterial site. Bioengineering Conference, 2002. Proceedings of the IEEE 28th Annual Northeast , 20-21 April 2002. Pages:61 - 62
Jandre F.C. Wavelet Analysis of Phonocardiograms: Differences between Normal and Abnormal Sounds. IEEE/EMBS 1997. pp. 1642-1644.
Johanson M. A Remote Auscultation Tool for Advanced Home Healthcare.
Kim, I.Y.; Lee, S.M.; Yeo, H.S.; Han, W.T.; Hong, S.H. Feature extraction for heart sound recognition based on time-frequency analysis. [Engineering in Medicine and Biology, 1999. 21st Annual Conf. and the 1999 Annual Fall Meeting of the Biomedical Engineering Soc.] BMES/EMBS Conference, 1999. Proceedings of the First Joint , Volume: 2 , 13-16 Oct. 1999. Pages:960 vol.2
Kim D, Tavel ME. Assessment of severity of aortic stenosis through time-frequency analysis of murmur. Chest (Chest.) 2003 Nov; 124(5): 1638-44
Kurnaz MN, Olmez T. Determination of Features for Heart Sounds by Using Wavelet Transforms. Computer-Based Medical Systems, 2002. (CBMS 2002). Proceedings of the 15th IEEE Symposium on , 4-7 June 2002 Page(s): 257 -261
Wavelet transform is applied to a window of two periods of heart sounds. Two analyses are realized for the signals in the window: segmentation of first and second heart sounds, and the extraction of the features. After segmentation, feature vectors are formed by using he wavelet detail coefficients at the sixth decomposition level. The best feature elements are analyzed by using dynamic programming. Seven categories of heart sounds were investigated.
Lewkowicz M and Gitterman M. Theory of heart sounds. Journal of Sound and Vibration, Volume 117, Issue 2, 8 September 1987, Pages 263-275.
Liang H, Hartimo J. A Heart Sound Feature Extraction Algorithm Based on Wavelet Decomposition and Reconstruction. IEEE EMBS, Vol. 20, No 3, 1998.
Liang H, Lukkarinen S, Hartimo I. A Heart Sound Segmentation Algorithm Using Wavelet Decomposition and Reconstruction. IEEE/EMBS 1997 pp. 1630-1633.
This study developed an algorithm based on the wavelet decomposition and reconstruction method to extract features from the heart sound recordings. An artificial neural network classification method was used to classify the heart sound signals into physiological and pathological murmurs. The classification result indicated 74.4% accuracy. The original recordings were segmented into four parts: the first heart sound, the systolic period, the second heart sound, and the diastolic period. The following features were extracted and used in the classification algorithm: a) Peak intensity, peak timing, and the duration of the first heart sound b) the duration of the second heart sound c) peak intensity of the aortic component of S2(A2) and the pulmonic component of S2 (P2) , the splitting interval and the reverse flag of A2 and P2, and the timing of A2 d) the duration, the three largest frequency components of the systolic signal and the shape of the envelope of systolic murmur e) the duration the three largest frequency components of the diastolic signal and the shape of the envelope of the diastolic murmur. The classifier used was a feed forwardz artificial neural network using the backpropagation learning method. Training set of 39 sound files (20 innocent murmurs), and testing set size of 39 sound files (20 innocent murmurs).
Liang H, Lukkarinen S, and Hartmo I. Heart Sound Segmentation Algorithm Based on Heart Sound Envelogram. IEEE Computers in Cardiology, Vol. 24, 1997 pp. 105-108.
Livanos, G.; Ranganathan, N.; Jiang, J. Heart sound analysis using the S transform. Computers in Cardiology 2000. pp. 587-590.
Leung TS. et.al. Analysing Paediatric Heart Murmurs with Discriminant Analysis. IEEE Conference in Medicine and Biology Society. Vol. 20, No. 3, 1998.
Leiserson, C.E. MURMUR CLINIC: AN AUSCULTATION EXPERT SYSTEM. MIT Laboratory for Computer Science. 1-1-1987
Leung TS, et.al. Analysis of the Second Heart Sounds for Diagnosis of Paediatric Heart Disease. IEE Proc.-Sci. Meas. Technol., Vol. 145, No. 6, Nov 1998.
Leung TS, et.al. Characterization of Paediatric Heart Murmurs Using Self-Organising Map. BMES/EMBS 1999. pp. 926.
In this paper pediatric heart murmurs are characterized using self-organized maps (SOM). Features are extracted from the time-frequency representations. SOM type of unsupervised artificial neural network.
Leung TS, et.al. Classification of heart sounds using time-frequency method and artificial neural networks. EMBS 2000 pp. 988-991.
In this study digitally recorded pathological and non-pathological phonocardiograms (PCGs) were characterized by the time-frequency method trimmed mean spectrogram. The heart sounds were recorded using an electronic stethoscope, Bosch. The features extracted from the TMS were the distribution of the systolic and diastolic signatures in the TF domain. In addition the acoustic intensities in systole and diastole were used as features. These features were used as inputs into the probability neural network. The results: sensitivity of 97.3%, and a specificity of 94.4%. The probability neural network contained 56 neurons and was implemented in MatLab.
Malarvili MB, Kamarulafizam I, Hussain S, Helmi D. Heart Sound Segmentation Algorithm Based on Instantaneous Energy of Electrocardiogram. Cpomputers in Cardiology 2003. pp. 327- 330.
Messer, Sheila. Cutting the Noise Out of Heartbeats. Science Daily. 08-30-2000.
Mgdob HM, Torry JN, Vincent R, Al-Naami B. The Application of the Morlet Transform Wavelet in the Detection of Paradoxical Splitting of S2. Computers in Cardiology 2003. pp. 323- 326.
Mohamed, A.S.A.; Raafat, H.M. Recognition of heart sounds and murmurs for cardiac diagnosis. Pattern Recognition, 1988., 9th International Conference on , 14-17 Nov. 1988 Pages:1009 - 1011 vol.2.
Myint WW, Dillard B. An Electronic Stethoscope with Diagnosis Capabilities. IEEE 2001.
Noponen A-L, Lukkarinen S, et.al. How to Recognize the Innocent Vibratory Murmur. IEEE Computers in Cardiology. 2000; 27: 561-564.
The goal of this study was to attempt to distinguish the innocent vibratory murmur from other systolic murmurs. The heart sounds of 807 children were digitized using an electronic stethoscope. The method used to analyze the heart signals was to replay the recorded signals and view the quantified timing, duration, and frequency contents of the murmurs in the spectrogram and phonocardiogram. In all of the 310 cases diagnosed with a vibratory murmur the spectrogram showed a well-defined are at 150 Hz range different from other systolic murmurs.
Nygaard H, Thuesen L, Terp K, Hasenkam JM, PaulSens PK. Assessing the severity of aortic valve stenosis by spectral analysis of cardiac murmurs (spectral vibrocardiography). Part I: Technical aspects. J Heart Valve Dis (The Journal of heart valve disease.) 1993 Jul; 2(4): 454-67.
Nygaard H, Thuesen L, Terp K, Hasenkam JM, PaulSens PK. Assessing the severity of aortic valve stenosis by spectral analysis of cardiac murmurs (spectral vibrocardiography). Part II: Clinical aspects. J Heart Valve Dis (The Journal of heart valve disease.) 1993 Jul; 2(4): 468-75.
Carroll JD ; Hellman KE. In response to: Nygaard H, Thuesen L, Hasenkam JM, Pedersen EM. Assessing the severity of aortic valve stenosis by spectral analysis of cardiac murmurs (spectral vibrocardiography) Part I and Part II. J Heart Valve Dis (The Journal of heart valve disease.) 1993 Sep; 2(5): 612-5.
Okada M. Chest wall maps of heart sounds and murmurs. Computers and Biomedical Research, Volume 15, Issue 3, June 1982, Pages 281-294.
Oskiper T, and Watrous W. Detection of the First Heart Sound Using a Time-delay Neural Network. IEEE Computers in Cardiology. 2002;29”537-540.
This work reports a method for detecting the first heart sound (S1) using a time-delayed neural network (TDNN). The network consists of a single hidden layer, with time-delayed links connecting the hidden units to the time-frequency energy coefficients of a Morlet wavelet decomposition of the input phonocardiogram (PCG) signal. The neural network operates on a 200 msec sliding window with each time-delay hidden unit spanning 100 msec of wavelet data. Heart sounds were recorded from 30 subjects for 20 seconds at each of four standard auscultatory sites using a commercially available electronic stethoscope. A training set comprised of half of the heartbeats from 20 randomly selected subjects was created. The network was trained on this set and tested on the full data set. The average performance was 1.6% deletion error and 2.2% insertion error. This level of S1 detection is considered satisfactory for analysis of the phonocardiogram signal.
Oskiper, T. and Watrous W. Results on the Time-Frequency Characterization of the First Heart Sound in Normal Man. Proceedings of the Second Joint EMSB/BMES Conference. 2002. pp. 126-127.
Ramnarayan P; Britto J. Paediatric clinical decision support systems. Arch Dis Child (Archives of disease in childhood.) 2002 Nov; 87(5): 361-2.
Reed TR, Reed NE, and Fritzson P. The Analysis of Heart Sounds for Symptom Detection and Machine-Aided Diagnosis. Eurosim 2001.
This work combined local signal analysis methods with classification techniques to detect, characterize, and interpret sounds corresponding to symptoms important for cardiac diagnosis. The system was evaluated using heart sounds corresponding to five different heart conditions. Heart sounds were hand segmented into a segment of a single heart beat cycle. Each segment was then transformed using 7 level wavelet decomposition, based on Coifman 4th order wavelet kernel. The resulting vectors 4096 values, were reduced to 256 element feature vectors, this simplified the neural network and reduced noise. The magnitude of the remaining coefficients in each vector were calculated and then normalized by the vector’s energy. Each vector was then classified using a three layer neural network (256 inputs nodes, 50 hidden nodes, and 5 output nodes). The training examples were shifted to provide shifting invariance, as wavelet decomposition is not generally shift invariant. For variances up to and including 3000 classification is 100% accurate for all heart sounds. Above a variance of 3000, the decrease in accuracy varies widely between the different sounds.
Reed TR, Reed NE, Fritzson P. Heart sound analysis for symptom detection and computer-aided diagnosis. Simulation Modelling Practice and Theory, Volume 12, Issue 2, May 2004, Pages 129-146
Reske D, Moussavi Z. Design of a Web-Based Remote Heart-Monitoring System. IEEE EMBS/BMES 2002. pp. 1847-1848.
Sharif, Z.; Daliman, S.; Sha'ameri, A.Z.; Salleh, S.H.S. An expert system approach for classification of heart sounds and murmurs. Signal Processing and its Applications, Sixth International, Symposium on. 2001 , Volume: 2 , 13-16 Aug. 2001. Pages:739 - 740 vol.2
Sakari, L.; Kari, S.; Anna-Leena, N.; Anna, A.; Raimo, S. Novel software for real-time processing of phonocardiographic signal. Engineering in Medicine and Biology society, 1997. Proceedings of the 19th Annual International Conference of the IEEE , Volume: 4 , 30 Oct.-2 Nov. 1997
Pages:1455 - 1457 vol.4
Say O, Dokur Z, Olmez T. Classification of Heart Sounds by using Wavelet Transform. EMBS/BMES Conference 2002, pp. 128-129.
Feature vectors are formed by using the wavelet detail and approximation coefficients at the second and sixth decomposition levels. The classification (decision making) is performed in 4 steps: segmentation of the first and second heart sounds, normalization process, feature extraction, and classification by the artificial neural network. Nine different heart sounds are classified.
Statsis A Ch, Loukis EN, Pavlopoulos, Koutsouris D. A Decision Tree-Based Method, Using Auscultation Findings, for the Differential Diagnosis of Aortic Stenosis from Mitral Regurgitation. Computers in Cardiology, 2003. pp. 769- 772.
Stasis A Ch, et.al. Using Decision Tree Algorithms as a Basis for a Heart Sound Daignosis Decision Support System. IEEE Conf on Information Technology Applications in Biomedicine 2003 pp. 354-357.
The first goal of this work was to evaluate the suitability of various decision tree structures, fully expanded Decision Tree Structures, and pruned Decision Tree Structures, for the given diagnostic subproblems. These were (a) distinguishing the Aortic Stenosis (AS) from the Mitral Regurgitation (MR) and (b) distinguishing the Opening Snap (OS), the Second Heart Sound Split (A2_P2) and the Third Heart Sound (S3). The second goal of this work was to evaluate the diagnostic abilities of the investigated heart sound features for Decision Tree based diagnosis. The heart sound signals were pre-processed to detect the first and second heart sounds in the following steps: a) wavelet decomposition, b) calculation of normalized average Shannon Energy, c) a morphological transform action that amplifies the sharp peaks and attenuates the broad ones d) a method that selects and recovers the peaks corresponding to S1 and S2 and rejects others e) algorithm that determines the boundaries of S1 and S2 in each heart cycle f) a method that distinguishes S1 from S2. The standard deviation of the duration of all heart cycles was used as the first four scalar features (F1-F4). The rest of the features were the two mean signals for each of the four structural components of the heart cycle. Heart sounds used in this study were collected from educational audiocassettes, audio CDs and CD ROMs. Results: discrimination of AS from MR was 88%, discrimination of A2_P2, OS, and S3 was 68.5%. Software used for decision tress: Envisioner.
Syed ZH, Guttag J, and Curtis D. Analyzing Heart Sounds. MIT Laboratory for Computer Science, March 2003.
Tanishiro, H.; Funakubo, A.; Kawamura, T.; Fukui, Y. Analysis of sound characteristics originating from the brachial artery in auscultation. [Engineering in Medicine and Biology, 1999. 21st Annual Conf. and the 1999 Annual Fall Meeting of the Biomedical Engineering Soc.] BMES/EMBS Conference, 1999. Proceedings of the First Joint ,Volume: 2 , 13-16 Oct. 1999
Thompson WR, Hayek CS, et.al. Automated Cardiac Auscultation for Detection of Pathologic Heart Murmurs. Pediatr Cardiol 22:373-379, 2001.
Torry JN. Heart sound analysis comparing wavelet and autoregressive techniques. Computers in Cardiology 2003. pp. 657- 660.
Tovar-Corona, B.; Hind, M.D.; Torry, J.N.; Vincent, R. Effects of respiration on heart sounds using time-frequency analysis. Computers in Cardiology 2001. pp. 457-460
Tovar-Corona, B.; Torry, J.N. Time-frequency representation of systolic murmurs using wavelets. Computers in Cardiology 1998 , 13-16 Sept. 1998
Pages:601 - 604.
Tovar-Corona, B.; Torry, J.N. Graphical representation of heart sounds and murmurs. Computers in Cardiology 1997 , 7-10 Sept. 1997
Pages:101 - 104.
Turkoglu I, Arslan A. An Intelligent Pattern Recognition System Based on Neural Network and Wavelet Decomposition for Interpretation of Heart Sounds. EMBS Conference, 2001 pp. 1747-1750.
Turkoglu I, Arslan A, Erdogan I. An intelligent system for diagnosis of the heart valve diseases with wavelet packet neural networks. Computers in Biology and Medicine, Volume 33, Issue 4, July 2003, Pages 319-331.
Varady, P. Wavelet-based adaptive denoising of phonocardiographic records. Engineering in Medicine and Biology Society, 2001. Proceedings of the 23rd Annual International Conference of the IEEE , Volume: 2 , 25-28 Oct. 2001. Pages:1846 - 1849 vol.2.
Watrous R, Reichek N. Siemens Corp. Multi-Modal Cardiac Diagnostic Decision Support System and Method. US Patent Application 20030093003.
White PR, Collis WB, Salmon AP. Time-Frequency of Heart Murmurs in Children. IEE Colloquium 3/1-3/4 1997.
Xin T. and Zhong T. Analysis and decision of heart sounds via ARMA models. Measurement, Volume 5, Issue 3, July-September 1987, Pages 102-106
Yoshida, H. Instantaneous Frequency Analysis of Systolic Murmur for Phonocardiogram. IEEE/EMSB 1997. pp. 1645-1647.
Zamri MZ, et.al. Wavelet Analysis and Classification of Mitral Regurgitation and Normal Heart Sounds Based on Artificial Neural Networks. IEEE pp. 619-620.
The application of wavelet transform for the heart sounds signal is described. The performance of integral wavelet transform and discrete wavelet transform for heart sounds analysis was discussed. The features from heart sounds were obtained from integral wavelet transform and used to train and test the artificial neural networks. The ANN was trained by 125 training data and tested with 52 data, with a classification accuracy of 94.2%. The heart sound data was recorded using an electronic stethoscope model FS203. The data consisted of 102 data from 22 patients with MR disease, and 75 data from 5 patients with healthy hearts. A multilayer perceptron neural network with an input, one hidden and an output layer were used to classify the heart sounds, with a maximum classification rate of 94.28%.
Murmur Clinic, MIT.
CADI - Cardiac Auscultation Diagnosis Instruction