基于极限学习机的室性早搏判别算法的实现  

Implementation of the Algorithm for Premature Ventricular Contraction Discrimination Based on Extreme Learning Machine

作　　者：王瑞荣[1] 余小庆[1] 王敏 叶杨 

机构地区：[1]杭州电子科技大学生命信息与仪器工程学院,杭州310018 [2]杭州红十字会医院骨科,杭州310003 [3]杭州红十字会医院医务科,杭州310003

出　　处：《中国生物医学工程学报》2017年第2期158-164,共7页Chinese Journal of Biomedical Engineering

基　　金：国家自然科学基金(61374005)

摘　　要：室性早搏是常见的心律异常疾病,给人的生命带来威胁,准确的心律异常诊断对于帮助人们预防心血管疾病起到重要的作用。以MIT-BIH心律异常数据库中的数据作为分析对象,提出一种基于极限学习机算法的诊断方法,主要包括信号预处理、特征提取和分类,实现心电信号室性早搏异常的判别。采用小波变换结合形态学算法,对信号进行预处理,去除干扰,得到纯净的心电信号。通过K-means聚类算法提取QRS波群等特征参数,根据这些参数建立正常窦性心律和室性早搏的正样本和预测样本,再结合极限学习机分类器进行样本训练匹配和分类识别。选取1 260个周期信号进行实验,结果表明,该算法能准确诊断出室性早搏异常,最终阳性平均检测率达到95%,平均灵敏度达到96%。该算法相比其他算法,在识别精度相当的情况下,可极大提高算法的实时性,具有很高的研究价值,同时在移动医疗和临床医疗方面也具有一定的实用价值。Premature ventricular contraction （PVC） is a common heart rhythm disorders, which threatens humanity＇ s health, therefore accurate diagnosis of abnormal heart rhythms plays an important role to help humanity prevent cardiovascular disease. This paper proposed a diagnosis method based on ELM （extreme learning machine, ELM） to realize the discrimination of PVC from normal ECG （electrocardiograph） using the data from the MIT-BIH database as analysis object, and process of the method includes signal preprocessing, feature extraction and classification. The first step was to apply the wavelet transform combined with morphological filtering method for signal preprocessing to get the relatively clean signal, Then extracted feature parameters of QRS complex by using K-means clustering algorithm. Meanwhile, the calibration samples and prediction samples were established according to the feature parameters, and finally the ELM classifier for sample training match and classification recognition was adopted. 1260 cycles of signal were chosen to do experiment, and the results demonstrated that this algorithm could accurately diagnose the PVC, whose positive detection rate was up to 95% and sensitivity was up to 96% on average. Compared with other algorithms in the condition of similar detection accuracy, this algorithm can improve the real-time performance of the algorithm, which has high research value and certain practical value in mobile medical treatment and clinical medicaltreatment.

关 键 词：极限学习机 室性早搏 心电信号 MIT-BIH 

分 类 号：R318[医药卫生—生物医学工程]