Robust Heart Sounds Segmentation Based on the Variation of the Phonocardiogram Curve Length

Authors: Mecheri Zeid Belmecheri, Maamar Ahfir, Izzet Kale

Abstract:

Automatic cardiac auscultation is still a subject of research in order to establish an objective diagnosis. Recorded heart sounds as Phonocardiogram (PCG) signals can be used for automatic segmentation into components that have clinical meanings. These are the first sound, S1, the second sound, S2, and the systolic and diastolic components, respectively. In this paper, an automatic method is proposed for the robust segmentation of heart sounds. This method is based on calculating an intermediate sawtooth-shaped signal from the length variation of the recorded PCG signal in the time domain and, using its positive derivative function that is a binary signal in training a Recurrent Neural Network (RNN). Results obtained in the context of a large database of recorded PCGs with their simultaneously recorded Electrocardiograms (ECGs) from different patients in clinical settings, including normal and abnormal subjects, show on average a segmentation testing performance average of 76% sensitivity and 94% specificity.

Keywords: Heart sounds, PCG segmentation, event detection, Recurrent Neural Networks, PCG curve length.

References:

[1] World Health Organisation, p. https://www.who.int/cardiovascular_diseases/en/.
[2] S. Choi and Z. Jiang, “Comparison of envelope extraction algorithm for cardiac sound signal segmentation,” Expert System, vol. 34, no. 2, pp. 1056, 1069, 2008.
[3] L. Sharma, “Multiscale Analysis of heart sound for segmentation using multiscale Hilbert envelope,” in 13 th International Conference on ICT and Knowledge, Bangkok, 2015.
[4] M. Ahfir and I. Kale, “Automatic Synchronization and Segmentation of Heart Sound Signals Based on their Continuous Average Energy,” Proc.of The 2nd World Congress on Multimedia and Computer Science, 2014.
[5] M. Z. Belmecheri, M. Ahfir and I. Kale, "Automatic heart sounds segmentation based on the correlation coefficients matrix for similar cardiac cycles identification," Biomedical Signal Processing and Control, vol. 43, pp. 300-310, May 2018.
[6] S. Schmidt, C. Holst-Hansen, C. Graff and E. Tofti, “Segmentation of heart sound recordings by a duration-dependent hidden Markov model,” Physiol Meas, vol. 31(4), pp. 513-529, 2010.
[7] A. Rowland and A. Sargent, The ECG workbook, M&K Publishing, 2011.
[8] C. Liu, D. Springer and G. Clifford, “"Performance of an open-source heart sound segmentation algorithm on eight independent databases,” Institute of Physics and Engineering in Medicin, vol. 38, p. 1730–1745, 2017.
[9] M. Beale, M. Hagan and H. Demuth, “Neural Network ToolboxTM User’s Guide,” in the Matworks inc, MA, USA, Natick, 2012.
[10] J. Elman, “Finding structure in time,” Cognitive Science, vol. 14, p. 179–211, 1990.
[11] “https://www.mathworks.com/help/deeplearning/ref/layrecnet.html,” (Online).
[12] Z. Olaofe, “A 5-day wind speed & power forecasts using a layer recurrent neural network (LRNN),” Sustainable Energy Technologies and Assessments, vol. 6, pp. 1-24, June 2014.
[13] D.B. Springer, “heart sound database,” https://physionet.org/physiotools/hss, 2016.
[14] L. Resnekov, “Understanding heart sounds and murmurs: with an introduction to lung sounds.,” JAMA, vol. 254, no. 1, p. 124–125, 1985.
[15] D. B. Springer, L. Tarassenko and G. D. Clifford, “Logistic Regression-HSMM-based Heart Sound Segmentation,” IEEE Transactions on Biomedical Engineering, vol. 63, no. 4, pp. 822-832, April 2016.