关键词: 相似性分析, 心电波形, 特征表示, 降维, 分类

Abstract: It is often difficult to classify Electrocardiogram （ECG） waveform automatically due to high similarity. A new feature representation of ECG waveform was proposed — window-slope method. In this method, an ECG waveform was divided into different windows in a plane, and the slope of maximum and minimum amplitude in a window was extracted as feature information to perform similarity analysis. The experimental results show that the method can not only reduce the dimension, but also can enlarge the difference between different types of waveforms under distance-based classification. The classification accuracy and efficiency can be improved by using the method; meanwhile the sensitivity and specificity of classification can be stabilized at a higher level.

Key words: similarity analysis, Electrocardiogram （ECG） waveform, feature representation, dimensionality reduction, classification