| 摘要: | 本篇論文包括“心電圖信號分析”及“心律不整的判斷”兩大研究主題。研究主題一是心電圖信號分析：我們提出一個簡單有效的方法稱之為差分運算法(Difference Operation Method (DOM))，它主要的目的是用來偵測心電圖(electrocardiogram (ECG)) 信號中的QRS 複合波。本主題分成下列的兩個執行步驟，分別是(1)將待測試的ECG信號作“差分運算”，其目的是要找R波；(2)以R波為基準，往R波的前面方向找Q波，往R波的後面方向找S波。以上的兩個步驟執行完後即可找到QRS 複合波，此時再繼續以現有的方法找P波與T波。最後我們以MIT-BIH 心律不整資料庫中的ECG信號來評估DOM的效能，經實際的測試，DOM比其它類似的方法有更高的正確判斷率，以及有更快的運算速度。研究主題二是心律不整的判斷：本主題我們提出了三種判斷心律不整的方法“Fuzzy Logic Method (FLM)”，“Linear Discriminant Analysis (LDA)”，及 “Fuzzy C-Means (FCM)”。這三種方法能判斷的心律包含正常的心跳 (normal heartbeats) 及不正常的心跳(abnormal heartbeats)。在本主題中，不正常的心跳包含“左束分支阻斷 (Left Bundle Branch Block)”、“右束分支阻斷(Right Bundle Branch Block)”、“心室過早收縮(Ventricular Premature Contractions)”及“心房過早收縮(Atrial Premature Contractions)”等四種較常發生的心律不整。我們以MIT-BIH 心律不整資料庫中的ECG信號來評估本主題中所提出三種判斷心律不整的方法之效能，經實際的測試，結果是：(1)處理30分鐘長的ECG信號 (約2100 次心跳) 需要的測試時間少於1分鐘(此時間包含主題一的心電圖信號分析時間)；(2)需要的記憶體空間大約是2 MB (2100 × 432 × 16 bits)；(3) FLM之正確判斷率是93.78%、 LDA 是96.23% 、FCM 是93.57%。依據以上的測試結果，結論是我們在本文中所提出的心電圖信號分析及心律不整的判斷是一個簡單有效及快速診斷的方法。 In this dissertation, some novel and efficient algorithms in two related research topics about ECG signals will be presented and discussed. In the first research topic, a simple and reliable method, called the Difference Operation Method (DOM), is proposed to detect the QRS complex of an electrocardiogram (ECG) signal. The proposed DOM includes two stages. The first stage is to find the point R by applying the difference equation operation to an ECG signal. The second stage looks for the points Q and S based on the point R and then finds the whole QRS complex. From the QRS complex, both T and P waves can be obtained by the existing methods. Some records (QRS complex and T, P waves) of ECG signals in MIT-BIH arrhythmia database are tested and thus it shows that the DOM has much more precise detection rate and faster speed than other existing methods. In the second research topic, three methods, named “Fuzzy Logic Method (FLM)”, “Linear Discriminant Analysis (LDA)”, and “Fuzzy C-Means (FCM)”, are applied for classifying the cardiac arrhythmia on ECG signals. The proposed methods can accurately classify the normal heartbeats and abnormal heartbeats. Abnormal heartbeats include Left Bundle Branch Block, Right Bundle Branch Block, Ventricular Premature Contractions and Atrial Premature Contractions. The proposed methods were evaluated using the MIT-BIH arrhythmia database and have the following advantages: (1) The average time required for processing 30-minute long records of ECG signals is less than 1 minute; (2) The maximum memory requirement is only about 2 MB (that is, 2100 × 432 × 16 bits) for 30-minute long (about 2100 beats) recordings with 16 bit sampling points; (3) Good detection results (High – Reliability). In the experiments, the average failure rate for processing 30-minute long records of ECG signals is 0.19% by DOM method. The total classification accuracy is 93.78%, 96.23% and 93.57% for FLM, LDA and FCM method, respectively. Thus, the proposed methods indeed provide efficient, simple and fast methods for classifying the cardiac arrhythmia from ECG signals. |
