邵燿華臺灣大學:應用力學研究所黃文彥Huang, Wen-YenWen-YenHuang2010-05-182018-06-292010-05-182018-06-292009U0001-0907200917554500http://ntur.lib.ntu.edu.tw//handle/246246/183533Abstractims 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. ethod 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 thisethod for classification.目錄文摘要 .......................................Ibstract .......................................II目錄..........................................III一章 序論......................................1-1 前言與研究動機...............................1-2 心電圖原理...................................2-3 MIT-BIH 心律不整資料庫.......................9-4 文獻回顧.....................................15-5 研究架構.....................................18二章 研究原理..................................19-1 經驗模態分解.................................19-2 動態時間扭曲演算法...........................24-3 主成分分析法.................................28-4 密度群聚.....................................31三章 實驗流程與結果............................38-1 特徵萃取.....................................39-2 主成分分析過程...............................41-3 以密度進行分群並驗證.........................43-4 分類結果的統計與分析.........................46四章 結果討論..................................51考文獻.........................................53application/pdf3702392 bytesapplication/pdfen-USECG本質模態函數(IMF)主成分分析(PCA)密度集群分析(DBSCAN)IMFPCADBSCANthesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/183533/1/ntu-98-R95543039-1.pdf