一种基于频谱包络特征提取的心音信号分类方法

Heart sound signal classification method based on spectral envelope feature extraction

摘要

针对心血管疾病易诱发心脏结构、功能异常，造成心肌收缩与血液泵血能力衰减、严重危及患者身心健康的问题，提出一种基于频率平衡功率谱强度特征的心音分析方法，可实现正常心音与病理性心音信号的有效区分。首先采用自适应小波阈值收缩降噪算法对心音信号进行预处理，提取其频率平衡功率谱强度（FBPSI）包络。再提取FBPSI包络的多维特征，融合功率谱密度、能量谱密度提取的特征，通过MRMR算法对特征进行重要性分数排序。最后应用机器学习模型对心音信号进行分类识别。实验采用Yaseen以及2016心音挑战赛（PhysioNet/CinC Challenge 2016）两个公开数据集进行验证。十折交叉验证结果表明，本研究提出的有效特征向量与K近邻分类器相结合在Yaseen数据集上5分类准确率为97.40%，在PhysioNet/CinC Challenge 2016数据集中正常/异常分类准确率为89.11%，可为心血管疾病的预防和诊断提供有效参考。

Abstract

Cardiovascular diseases can predispose the heart to structural and functional abnormalities, leading to weakened myocardial contractility and reduced blood-pumping capacity, which severely jeopardizes patients' physical and psychological health. Therefore, this study proposes a heart sound analysis method based on frequency-balanced power spectrum intensity (FBPSI) features to enable effective differentiation of normal and pathological heart sound signals. An adaptive wavelet threshold shrinkage denoising algorithm is first employed to preprocess heart sound signals and extract FBPSI envelopes. Subsequently, multi-dimensional features are extracted from the FBPSI envelopes, which are then fused with features derived from power spectral density and energy spectral density. The importance of these features is ranked using the minimum redundancy maximum relevance algorithm. Finally, machine learning models are applied for the classification and identification of heart sound signals. Experimental validation is conducted using the Yaseen and the PhysioNet/CinC Challenge 2016 public datasets. Ten-fold cross-validation results demonstrate that the proposed effective feature vectors combined with the K-nearest neighbor classifier achieves a five-class classification accuracy of 97.40% on the Yaseen dataset and a binary classification accuracy of 89.11% on the PhysioNet/CinC Challenge 2016 dataset, providing valuable references for the prevention and diagnosis of cardiovascular diseases.

关键词

心音分类 / 功率谱强度 / 特征提取 / 心脏瓣膜疾病 / 频域包络

Key words

heart sound classification / power spectrum intensity / feature extraction / heart valve disease / frequency domain envelope

参考文献

[1] 陈喆, 蒋薇薇, 朱佳敏, 等 . 全球重大慢性病专利演进态势及我国发展对策研究[J]. 中国发明与专利, 2025, 22(3): 65-73.

[2] Chen Z, Jiang WW, Zhu JM, et al. Analysis of patent development trend of major chronic diseases and development strategies in China[J]. China Invention & Patent, 2025, 22(3): 65-73.

[3] Jneid H, Chikwe J, Arnold SV, et al. 2024 ACC/AHA clinical performance and quality measures for adults with valvular and structural heart disease: a report of the American Heart Association/American College of Cardiology Joint Committee on Performance Measures[J]. Circ Cardiovasc Qual Outcomes, 2024, 17(4): e000129.

[4] 王志, 张菲菲, 陈始圆, 等 . 心音智能听诊技术的研究现状与应用展望[J]. 软件导刊, 2023, 22(7): 212-219.

[5] Wang Z, Zhang FF, Chen SY, et al. Research status and application prospect of intelligent auscultation technology of heart sound[J]. Software Guide, 2023, 22(7): 212-219.

[6] Nizam NB, Nuhash SI, Hasan T . Hilbert—envelope features for cardiac disease classification from noisy phonocardiograms[J]. Biomed Signal Process Control, 2022, 78: 103864.

[7] Oh SL, Jahmunah V, Ooi CP, et al. Classification of heart sound signals using a novel deep WaveNet model[J]. Comput Methods Programs Biomed, 2020, 196: 105604.

[8] Avanzato R, Beritelli F . Heart sound multiclass analysis based on raw data and convolutional neural network[J]. IEEE Sens Lett, 2020, 4(12): 1-4.

[9] 房玉, 昌业勤, 郭子健, 等 . 基于小波包重构信号能量分布特征的心音分类识别[J]. 中国医学物理学杂志, 2024, 41(2): 205-211.

[10] Fang Y, Chang YQ, Guo ZJ, et al. Heart sound classification using energy distribution features extracted with wavelet packet decomposition[J]. Chinese Journal of Medical Physics, 2024, 41(2): 205-211.

[11] 罗刚, 王铭勋, 黎明, 等 . 面向情绪脑电分析的增强型功率谱密度特征提取方法[J]. 中国医学物理学杂志, 2022, 39(3): 349-356.

[12] Luo G, Wang MX, Li M, et al. Feature extraction method based on enhanced power spectral density for emotion analysis using EEG[J]. Chinese Journal of Medical Physics, 2022, 39(3): 349-356.

[13] 杨利英, 孟天昊, 张清杨, 等 . 特征融合实现脑电信号情感分析[J]. 西安电子科技大学学报, 2022, 49(6): 95-102.

[14] Yang LY, Meng TH, Zhang QY, et al. Implementation of EEG emotion analysis via feature fusion[J]. Journal of Xidian University, 2022, 49(6): 95-102.

[15] 许春冬, 辛鹏丽, 周静, 等 . 基于功率谱密度与卷积神经网络的心音分类[J]. 计算机工程与应用, 2021, 57(10): 125-132.

[16] Xu CD, Xin PL, Zhou J, et al. Classification of heart sounds using power spectral density and convolutional neural networks[J]. Computer Engineering and Applications, 2021, 57(10): 125-132.

[17] Riccio D, Brancati N, Sannino G, et al. CNN—based classification of phonocardiograms using fractal techniques[J]. Biomed Signal Process Control, 2023, 86(Part B): 105186.

[18] 李伟, 杨向东, 陈恳 . 基于CNN和RNN联合网络的心音自动分类[J]. 计算机工程与设计, 2020, 41(1): 46-51.

[19] Li W, Yang XD, Chen K . Heart sound classification based on CNN and RNN[J]. Computer Engineering and Design, 2020, 41(1): 46-51.

[20] Li F, Tang H, Shang S, et al. Classification of heart sounds using convolutional neural network[J]. Appl Sci, 2020, 10(11): 3956.

[21] Yaseen, Son GY, Kwon S . Classification of heart sound signal using multiple features[J]. Appl Sci, 2018, 8(12): 2344.

[22] Liu CY, Springer D, Li Q, et al. An open access database for the evaluation of heart sound algorithms[J]. Physiol Meas, 2016, 37(12): 2181-2213.

[23] 王燕, 王海滨, 刘立汉 . 基于小波变换的心音信号降噪方法[J]. 信息与电子工程, 2010, 8(3): 303-307.

[24] Wang Y, Wang HB, Liu LH . Noise reduction for heart sound based on wavelet transform[J]. Information and Electronic Engineering, 2010, 8(3): 303-307.

[25] 汪琴, 杨宏波, 潘家华, 等 . 基于时频组合特征与自适应模糊神经网络的心音分类算法[J]. 生物医学工程学杂志, 2023, 40(6): 1152-1159.

[26] Wang Q, Yang HB, Pan JH, et al. Heart sound classification algorithm based on time—frequency combination feature and adaptive fuzzy neural network[J]. Journal of Biomedical Engineering, 2023, 40(6): 1152-1159.

[27] Sanjeet S, Sahoo BD, Parhi KK . Low—energy real FFT architectures and their applications to seizure prediction from EEG[J]. Analog Integr Circuits Signal Process, 2023, 114(3): 287-298.

[28] Jing H, Jie R, Siqi L, et al. A cardiac audio classification method based on image expression of multidimensional features[J]. Sci Rep, 2024, 14(1): 23196.

[29] Fang Y, Leng HX, Wang WB, et al. Multi—level feature encoding algorithm based on FBPSI for heart sound classification[J]. Sci Rep, 2024, 14(1): 29132.

[30] Orozco—Reyes L, Alonso—Arévalo MA, García—Canseco E, et al. A deep—learning approach to heart sound classification based on combined time—frequency representations[J]. Technologies, 2025, 13(4): 147.

[31] Boz C, Kocyigit Y . Segmented vs. non—segmented heart sound classification: impact of feature extraction and machine learning models[J]. Appl Sci, 2025, 15(20): 11047.

[32] Ravikiran HK, Deepak R, Deepak HA, et al. A robust framework for Alzheimer's disease detection and staging: incorporating multi—feature integration, MRMR feature selection, and Random Forest classification[J]. Multimed Tools Appl, 2025, 84(22): 24903-24931.

[33] Wang JL, Zang JB, Yao S, et al. Multiclassification for heart sound signals under multiple networks and multi—view feature[J]. Measurement, 2024, 225: 114022.

[34] Zhang XF, Liu XH, Liu GZ . A heart sound signal classification method based on the mixed characteristics of Mel cepstrum coefficient and second—order spectrum[J]. Circuits Syst Signal Process, 2024, 43(6): 3533-3552.

基金资助

国家自然科学青年基金(61901393)

湖南省教育厅科学研究项目(24C1214)