| Summary: | 碩士 === 健行科技大學 === 電子工程系碩士班 === 106 === This study has two major themes. The first theme is to design an Electrocardiogram (ECG) signal analyzer with Weighted Principal Component Analysis (WPCA) algorithm. The second theme is to determine heartbeat case by using the proposed ECG signal analyzer. The ECG signal analyzer consists of two concatenation levels. The first level is the pre-signal processing level which is responsible for noise removal and signal amplification of the patient''s original ECG signal. The second level is the signal analysis level which can analyzes ECG signals using WPCA algorithm. This level is composed of the following three procedures, namely: (i) Procedure-FWV: The main purpose is to calculate final weights values of the original component (or original feature points) in different clusters; (ii) Procedure-DPC: The main purpose is to determine the number of the final principle components or the qualitative features in different clusters; and (iii) Procedure-CCD: The main purpose is to determine the final cluster of the ECG signal to be tested.
In addition, the second theme is to apply the ECG signal analyzer to determine heartbeat types. This study identifies six types of heartbeats, including normal heartbeat (NORM) and five abnormal heartbeats (LBBB, RBBB, VPC, APC, PB, etc.). Finally, in order to assess the performance of the ECG signal analyzer presented in this study, we use the relevant files in the MIT-BIH Arrhythmia Database to evaluate the effectiveness of the proposed method. After several tests, the test results are as follows: The sensitivity of the heartbeat classes NORM, LBBB, RBBB, VPC, APC, and PB are 97.80%, 91.54%, 93.53%, 90.29%, 89.66%, and 84.25%, respectively. The total classification accuracy is 94.05%.
|
|---|
