心音信号的特征矢量辨识分析

Analysis of PCG characteristic vector recognition

The authors have estimated the coefficient of PCG AR (p) model to be the sample characteristic vector by the Burg algorithm and the theory of the steady random process. One PCG recognition area has been deter mined according to the coefficient. In the present study, 6 cardiac cycles of 3 normal subjects and 4 cardiac cycles of 2 patients with heart diseases were recorded by multi- channel physiological recorder in the clinical trial of the recognition method. The S1 and S2 recognition areas of 6 cardiac cycles of 3 normal subjects were employed as discrimination standard so as to compare and analyze the recognition results of 2 patients with heart diseases. The results showed that in the 5 participants, the S1 recognition area of 3 normal subjects and 2 patients with heart diseases overlapped, though the S2 area did not. This indicated that the S, of 2 patients was normal while the S2 was abnormal. These findings also agree with the physiological implications of sampled experimental data.