基于心动周期估计的心音分割及异常心音筛查算法

发布时间：2018-07-07 08:18

本文选题：心音分割 + 异常心音筛查　；参考：《电子与信息学报》2017年11期

【摘要】：心脏疾病是全球发病率和死亡率最高的疾病,心音听诊可以获取心脏的机械特性及结构特征,与超声心动图、核磁共振等无创诊断技术相比具有快速、低成本和操作简单的优势。心音信号成分复杂,容易受到各种噪声和干扰的影响,听诊诊断结果容易受到医生主观性的影响,极大限制了心音听诊的应用。该文提出一种基于心动周期估计的心音分割及异常心音筛查算法,预先估计了心音的心动周期,存在随机干扰的情况下也可以正确识别信号中80%以上的心动周期,提高了算法的稳定性。同时提出了区分度良好的时域和频域特征指标,利用支持向量机建模,对异常心音的识别率可达92%。算法可辅助医生诊断,或用于家用便携式心音监护设备。
[Abstract]:Heart disease has the highest morbidity and mortality in the world. Heart-tone auscultation can obtain the mechanical and structural characteristics of the heart, which is faster than non-invasive diagnostic techniques such as echocardiography and MRI. The advantages of low cost and simple operation. The composition of heart sound signal is complex and easily affected by various noises and disturbances. The diagnosis result of auscultation is easily influenced by the subjective of the doctor, which greatly limits the application of auscultation of heart sound. In this paper, an algorithm of cardiac sound segmentation and abnormal heart sound screening based on cardiac cycle estimation is proposed. The cardiac cycle of heart sound can be estimated in advance, and more than 80% of the cardiac cycle in the signal can be correctly identified in the presence of random interference. The stability of the algorithm is improved. At the same time, the time-domain and frequency-domain characteristic indexes with good discrimination are proposed. The recognition rate of abnormal heart sounds can reach 92% by using support vector machine (SVM) modeling. The algorithm can assist the doctor in diagnosis, or be used in portable heart sound monitoring equipment.
【作者单位】：中国科学院电子学研究所;中国科学院大学;中国人民解放军海军总医院;
【基金】：国家自然科学基金(61302033) 北京市自然科学基金(Z160003) 国家重点研发计划(2016YFC1304302,2016YFC0206502,2016YFC1303900)~~
【分类号】：R540.4;TN911.7

【相似文献】