Measurement 12-lead ECG Character Points Using Line Segment Clustering Technique

• Main Authors: Shey Jenn-taur, 徐震濤
• Other Authors: Walter H. Chang
• Format: Others
• Language: zh-TW
• Published: 1998
• Online Access: http://ndltd.ncl.edu.tw/handle/14893385544840934290

碩士 === 中原大學 === 醫學工程研究所 === 86 === A new technique for ECG character points measurement is presented in this study. Linear Approximation Distance Thresholding (LADT) and Line Segment Clustering (LSC) technique are combined to process ECG signal and to recognize the important points of one beat ECG signal. In this study, several character points, such as R peak and T peak are recognized by this technique, and the other, PR, QRS, QT interval are measured by this system. The results of LSC technique are compared to that of the manual ones. Simulation signals which combine with different SNR levels (5%, 10%, 15%, 20%, 25%, 30% ) white noise and from 10 healthy volunteer 12-lead ECG signals were used to evaluate this system. The error for noisy simulation signal (SNR:30%) is less than 7 points. The error for real signal is less than 4 points. The error of interval value, compared to the other instrument, are less then 0.1 sec; compared to manual result, the error are less than 0.05 sec.