心電圖(ECG)分析是檢測心律不整最好的方法之一,現今也有著許多相關的演算法被提出。本論文分成兩個部份,演算法實驗部分將使用支持向量機 (Support Vector Machines)整合MIT-BIH Arrhythmia database,與心電圖檢視系統。目標是利用小波轉換與其他生理特徵,來抽取心電圖特徵並以MIT-BIH Arrhythmia database當作支持向量機的訓練基底模組,其中將資料切割成兩個部分,一部分拿來做訓練、另一部分拿來做測試,並且得到了98%的高正確率。 最後實做部分使用心電圖檢視系統,利用廠商合作的心電圖量測儀經由藍芽將資料存在客戶端,並且利用Web Service將讀入量測的心電圖資料連結到支持向量機系統,並且分析出心電圖資料狀況,即時的得知分析結果回傳到網頁上顯示。
Electrocardiogram (ECG) analysis is one of the best ways to detect cardiac arrhythmias, today also number of related algorithms have been proposed [20-23]. This thesis is divided into two parts, algorithm experiment and system implementation. We will use the Support Vector Machine (SVM) and MIT-BIH Arrhythmia database in an EGC monitor system. Our goal is to use wavelet transform and other physiological characteristics to extract ECG features, and the Arrhythmia database is used as the training module of Support Vector Machine. The next, web service will link to Support Vector Machine when we use EGC monitor view system to read the ECG data. The status of ECG data will be analyzed and the result returned to the healthcare provider in real-time.