題名: A Pathological ECG Recognition System Using Principal Component Analysis Neural Networks
作者: 余松年;楊朝樑
貢獻者: 中正大學電機系;新民高中資訊科
關鍵詞: 心電圖;主要元素分析;資料空間;特徵空間;ECG;PCA;data space;feature space
日期: 2003-09-13
上傳時間: 2009-12-09T05:25:07Z 出版者: 臺中健康暨管理學院
摘要: 本研究之心電圖(Electrocardiogram, ECG)資料來自 MIT/BIH 資料庫[1]，以 QRS 複合波之 R 波為中心，共取 64 點心電圖資料，直接輸入至主要元素分析 (Principal Component Analysis,PCA)類神經網路[2][3]，將輸入資料由資料 空間(dataspace)轉換成特徵空間(feature space)，以取得輸入資料特徵值，
再利用最小歐氏距離(Euclidean distance)平方做辨識，辨識率可達 98% 以 上。
The ECG data used for this study came from the MIT/BIH database.
The data were centered at the R wave of the QRS complex and 64 points were used. The data were used as inputs to the principal component analysis (PCA) neural network. The PCA
neural network transforms the data into the principal features of the data. Euclidean distance was used to measure the similarities
between the clustered feature and the test data. The recognition rate of our system was found to be as high as 98%.
