Automatic Segmentation Algorithm of Heart Sounds Based on Its Periodicity
XU Lili, GUO Xueqian
Beijing Key Laboratory of Fundamental Research on Biomechanics in Clinical Application, School of Biomedical Engineering,
Capital Medical University, Beijing 100069, China
Abstract:Objective It is known that the automatic segmentation of cardiac voice signal is the first task of the normal and abnormal
recognition of the heart sounds. The present study aims to automatically segmente the heart sound signal by the periodic and
physiological characteristics of the heart sound signal. Methods Forty-eight cases of heart sounds including 10 cases of abnormal
were chosen to perform the analysis. Two heart sound signal samples of 5 s extracted for each case, and a total of 96 samples were
gained. Mathematical morphology and fragment related calculation algorithm were combined to automatic extract the heart sounds
signal envelope. Then, main peak segment was cut out for correlation with whole envelope to mark the start of each cardiac cycle. S1
and S2 were determined according to physiological characters of heart sounds. Results The proposed method could display normal
signals for 100% accuracy and 89% for abnormal signals. Conclusion This algorithm can be used to automatically segment the heart
sound, and can be used for the feature extraction of the sound recognition.
许莉莉,郭学谦. 基于心音周期性的自动分段研究[J]. 中国医疗设备, 2018, 33(1): 86-88.
XU Lili, GUO Xueqian. Automatic Segmentation Algorithm of Heart Sounds Based on Its Periodicity. China Medical Devices, 2018, 33(1): 86-88.