摘要
基于心音听诊急性心梗(AMI)病发即时响应的关键在于心音特征异常监测精准率,然而由于患者病发状态的非确定性导致基于心音成分分割准确度的特征提取难度较差。鉴于此,本研究提出一种新的基于短时修正希尔伯特变换(STMHT)的自适应分割方法(ASTMHT)实现心音自适应分割以确保提取特征的准确性。实验通过对比平均香浓能量(ASE)算法和STMHT算法在AMI临产数据以及不同状态心音数据集的分割精度分析。对比结果表明,ASTMHT方法识别效果优于平均香农能量ASE方法和STMHT方法,尤其是在非安静状态数据集上,分割准确率比平均香农能量提高了53.95%,比STMHT提高76.57%。进而可为AMI特征提取提供保障。
The key to immediate response in acute myocardial infarction (AMI) diagnosis through heart sound auscultation lies in the precision of abnormal heart sound feature detection. However, the uncertainty of patients’ conditions leads to challenges in achieving accurate feature extraction based on heart sound segmentation. To address this, this study proposes a novel adaptive segmentation method (ASTMHT) based on short-time modified Hilbert transform (STMHT) to ensure precise feature extraction through adaptive heart sound segmentation. Experimental analysis compared the segmentation accuracy of the average Shannon energy (ASE) algorithm and the STMHT algorithm on AMI clinical data and heart sound datasets under various conditions. Results demonstrate that the ASTMHT method outperforms both the ASE and STMHT methods, particularly on non-quiet state datasets, with segmentation accuracy improvements of 53.95% over ASE and 76.57% over STMHT, thereby providing a robust foundation for AMI feature extraction.
出处
《计算机科学与应用》
2025年第8期260-269,共10页
Computer Science and Application
基金
本研究获湖南省大学生创新创业重点项目(No. S202410543068)、教育厅重点项目(No. 21A0403, No. 21A0405)、湖南省自然科学基金项目(No. 2022JJ30282, No. 2023JJ30283)、教育部产学研合作项目(No. 202102211006, No. 230806093021436)基金项目资助。
