基于DBSCAN聚类算法的心电图R峰检测研究  

Research on electrocardiogram R peak detection based on DBSCAN clustering algorithm

作　　者：潘潇潇 郑建立[1] PAN Xiaoxiao;ZHENG Jianli(School of Health Science and Engineering,University of Shanghai for Science and Technology,Shanghai 200093,China)

机构地区：[1]上海理工大学健康科学与工程学院,上海200093

出　　处：《智能计算机与应用》2024年第7期198-202,共5页Intelligent Computer and Applications

摘　　要：本研究致力于提高包含运动噪声、肌电噪声和基线漂移等干扰的心电信号中R峰的检测准确性,为准确估计心率和心率变异性等重要生理参数提供依据。研究通过整合4种不同的R峰检测算法Pan_Tompkins、Hamilton、Engzee和GQRS的结果,并应用DBSCAN聚类算法,用于提升R峰的检测性能。实验选用布尔诺理工大学心电图质量数据库作为数据源,从中提取了1000 s的第二类信号质量样本,这些样本含有运动伪迹和噪音干扰,但R峰依旧可识别,R峰定位的误差容忍度设为100 ms以内。实验结果表明,DBSCAN算法显著提高了R峰检测的准确率至87.6%,灵敏度至93.1%,阳性预测值至93.8%,以及F分数至0.934,说明DBSCAN算法能有效提升在复杂噪音环境下的R峰检测性能,满足临床及研究需求。This study aims to improve the detection accuracy of R-peaks in electrocardiogram signals containing interference such as motion noise,electromyographic noise,and baseline drift,providing a basis for accurately estimating important physiological parameters such as heart rate and heart rate variability.The study aims to improve the detection performance of R peaks by integrating the results of four different R peak detection algorithms,Pan-Tompkins,Hamilton,Engzee,and GQRS,and applying the DBSCAN clustering algorithm.The experiment selected the electrocardiogram quality database of the University of Technology of Brno as the data source,and extracted 1000 seconds of second-class signal quality samples.These samples contain motion artifacts and noise interference,but the R-peak can still be recognized,and the error tolerance for R-peak positioning is set to be within 100 milliseconds.The experimental results show that the DBSCAN algorithm significantly improves the accuracy of R-peak detection to 87.6%,sensitivity to 93.1%,positive predictive value to 93.8%,and F-score to 0.934,indicating that the DBSCAN algorithm can effectively improve the performance of R-peak detection in complex noise environments,meeting clinical and research needs.

关 键 词：心电信号 R峰检测算法 DBSCAN 

分 类 号：R339.4[医药卫生—人体生理学]