摘要
利用ST段偏移时心率变异性(HRV)存在的变化,探讨区分ST偏移时段和正常时段,进而区分缺血或单纯心率改变引起ST段偏移的可行性。从Long-term ST-T(LTST)数据库免费下载的心电信号数据中,选取25个记录中缺血引起ST段偏移时段(105段)和单纯心率变化引起ST段偏移时段(43段),以及所选时段前后各5min作为对照时段。选用平滑伪Wigner-Ville分布(SPWVD),计算不同时域段和频域段的心率变异性(HRV)分析参数。利用Fisher线性判别和支持向量机进行判别分析,并采用留一法检验分类判别的正确率。对于Fisher线性判别,ST段偏移时段判别的正确率随时段持续时间的延长而减小,而不同诱因区分的正确率却较高(不低于89.7%)。支持向量机法对上述各种检测和分类判别的正确率均为100%。研究结果为从非形态学的角度检测ST段偏移时段的存在和产生原因的判别,提高心肌缺血判别的准确性,提供了有参考价值的信息。
In this study, time-frequency method based analysis of heart rate variability （HRV） was used to detect the existence of ST-segment deviation episodes and the classification of ischemic and heart-rate related episodes. From 25 free download ECG records in long-term ST-T （LTST） database, 105 ischemic ST-segment deviation episodes and 43 heart-rate related ones were selected. 5 min intervals before and after each selected episode were extracted as control ones. Smoothed pseudo Wigner Ville distribution （SPWVD） based HRV parameters were calculated at different temporal divisions and different frequency bands. Fisher linear discriminant analysis and support vector machine （SVM） were used to find out sensitive parameters for the detection of ST-segment deviation episodes and the classification of them. leave-one-out method was employed for the accuracy test. For Fisher linear discriminant analysis, the accuracy of identifying ST-segment deviation episodes decreased with the increase of episode duration. The accuracy was relatively high （no less than 89.7% ） for the classification of ischemic and heart-rate related episodes. SVM showed accuracy of 100% in all kinds of detection and classification. As a non-morphological method, HRV analysis based on time-frequency method is useful in identifying the presence of ST-segment deviation episodes and the classification of them. The investigation lays a basis for promoting the accuracy of myocardial ischemia detection.
出处
《中国生物医学工程学报》
CAS
CSCD
北大核心
2008年第1期23-28,共6页
Chinese Journal of Biomedical Engineering
基金
北京市自然科学基金(3052015)
