A Neuro-Network Approach to Measure ST-Segment
Jun Yang · Beijing shengwu yixue gongcheng · 2002
Measurement of ECG′s ST Segment plays an important role in routine ECG diagnosis. A method which uses Neuro Network to analyze ECG′s ST Segment is presented in this paper. A BP(Back Propagation) Neuro Network was trained to learn the known ECG′s ST Segment morphological shapes. After training, the system was improved to recognize different types of ST Segment, and identify the start point and the end point of the ST Segment more accurately. By this method, more types of ST Segment index measured can be done more accurately. “Standardi zation” of the ECG segment were done during preprocessing, which made the ST Segment measurement and morphological recognition essentially be not affected by heart rate and amplitude. It can be used to analyze of ECG data of people doing exercise. In ECG reports, we added ST Segment morphological classification, it may be helpful for doctors to analyze reasons of ST Segment variation. We also give details of the structure and the training methods of our BP Neuro Network, and we discuss the advantages and disadvantages of the method presnted.