论文部分内容阅读
针对传统心律失常自动分类算法大多利用线性变换方法只提取单一特征的问题,本文提出基于多特征融合和纠错编码支持向量机的识别算法。该算法利用核独立成分分析(KICA)提取心电(ECG)信号的非线性特征,利用小波分析提取时频域特征,二者融合形成多域特征能够更全面反映不同类型心律失常的模式。设计基于受试者工作特性ROC曲线下面积指标优化的纠错编码支持向量机分类器,该指标比传统正确率指标能够更好的评价分类器性能。针对MIT-BIH心律失常数据库数据的实验表明,所提出算法优于基于单一特征的传统方法,ROC曲线下面积值为0.956,具有很好的分类效果。
In order to solve the problem that the traditional automatic classification algorithm of arrhythmia uses only the linear transformation method to extract only one feature, this paper proposes a recognition algorithm based on multi-feature fusion and error-correcting coding support vector machine. The algorithm extracts the nonlinear features of ECG signals by kernel independent component analysis (KICA), extracts the features of time-frequency domain by using wavelet analysis, and combines the two to form multi-domain features to fully reflect the different types of arrhythmia patterns. An error-correcting coding SVM classifier based on the optimization of the area under ROC curve was designed, which can better evaluate the performance of the classifier than the traditional accuracy index. Experiments on MIT-BIH arrhythmia database data show that the proposed algorithm is superior to the traditional method based on single feature, the area under the ROC curve is 0.956, which has good classification results.