基于突变点检测与峰值搜索的心音分割算法

Heart Sound Segmentation Algorithm Based on Mutation Point Detection and Peak Search

心音分割通常指将心音信号的心动周期分割为第一心音(S1)、收缩期、第二心音(S2)和舒张期;心音分割是分析研究心音分类的关键步骤。提出一种新的心音分割算法,首先采用bior双正交小波对心音进行去噪;然后根据短时能量、频谱质心以及阈值定位心音信号突变点,并确定寻峰区间;之后用寻峰算法滤除伪峰的同时通过峰值点补偿算法提高有效峰值的检出率;最后,利用峰峰值之间的距离定位S1和S2,并以突变点为分割点,得到心音分段结果。实验证明,使用上述算法对1000例心音进行分段,平均准确率达到了92.38%,平均分割速度为0.02561秒,在保证准确率的前提下实现了对心音信号的快速分割。

Heart sound segmentation usually means that the heart cycle of heart sound signal is divided into first heart sound(S1),systole,second heart sound(S2),and diastole.Heart sound segmentation is a key step in the anal-ysis of heart sound classification.A novel algorithm of heart sound segmentation was proposed in this paper.First,Bior biorthogonal wavelet was used to denoise the heart sound.Then,mutation points of the heart sound signal were detected according to the short-time energy,spectral centroid,and the threshold.The peak-seeking intervals were also determined.Next,the peak-seeking algorithm was used to filter out the false peak.The detection rate of the ef-fective peak was improved through the peak point compensation algorithm.Finally,the distance between peak and peak was used to locate S1 and S2.The heart sound was segmented according the mutation points.1000 cases of heart sound were used in this study.Experiments show that an average accuracy of 92.38%,and the average segmentation speed of 0.02561 seconds are achieved using the novel method,so that fast segmentation of heart sound signal is real-ized under the premise of ensuring accuracy.