心電圖型態分類:應用本質模態特徵

Abstract

Abstract Aims Morphological classification of the single heartbeat is the most important part of the computer aided Arrhythmia Analysis. The operations of these systems applied can be divided into four steps:1. The removal of noise and artifacts ; 2. Fiducial points detection; 3. Morphological classification;4. The rhythm analysis and medical interpretation . In this paper, our aim was to classify the heartbeat into various groups. Method and results we use the method based on the Empirical Mode Decomposition algorithm and Dynamic Time Warping algorithm for extraction of features that can be used to classify various abnormal heartbeats. Further, we reduce the dimensionality of data in the form of n features of a vector with p variables used to principal component analysis .The performance of our algorithms has been evaluated by MIT-BIH Arrhythmia Database. According to the experimental result, the accuracy of all beats is approximately equal to or greater than 85% with the overall accuracy being 90%. This indicates the effectiveness of this method for classification.