Skip to Main Content
If life-style related diseases could not be monitored continuously during a long time some heart defects might be difficult to be diagnosed appropriately and detected in an early step. Furthermore, the need for the primary health care physicians to improve the cardiac auscultation skill is still very strong in the primary screening examination, and becomes stronger for the general users to perform the auscultation at home. The objective of this paper presents the novel detection method for heart defects using the cardiac sound characteristic waveform (CSCW) with data clustering technique. An analytical model based on a mass-spring-damper system is proposed for extracting the CSCW from the sound signals. Feature sets T1, T2, T1T1 and T1T2 induced from the time elapses of the first heart sound and second heart sound with a passive threshold value (THV) are, also, introduced to detect heart abnormalities. Further, data clustering technique will be introduced to determinate an adaptive and reliable THV ranges. The cost function, e.g., Jm, and two cluster centers, e.g., (c11, c12) and (c21, c22), of the feature sets are also used to identify normal and abnormal heart sounds automatically. The feature sets were verified useful for identification of normal and abnormal heart sounds. The easy-understanding graphical screening ways of the features was considered, in advance, even for an inexperienced user able to monitor his or her pathology progress. Furthermore, for clustering results, the minimized cost function and cluster centers could be also efficient indicators for identifying the heart defects. Finally, a case study on the normal heart sound and abnormal heart sound was demonstrated to validate the usefulness and efficiency of CSCW with clustering algorithm. Particularly, the normal cases had very small value. For abnormal cases, in case of aortic regurgitation, its Jm was very small and the values of- the centers were very high comparing to the normal cases.