Automatic Detection of Myocardial Ischemia in ECG Using ANN and ANFIS
Abstract
Myocardial Ischemia occurs when heart muscle does not receive enough oxygen supply, the early detection of Ischemia helps the treatment process and preventing the progress of myocardial infarction. The objective of this work is to find an automatic method to help Cardiologists detecting ischemic episodes in ECG in order to diagnose Myocardial Ischemia.
Data from the European Database ST-T was used in this research, the signals were preprocessed for noise reduction and QRS complexes detection, followed by feature selection using Principle Component Analysis (PCA). Finally, two types of classifiers based on Artificial Neural Network (ANN) and Adaptive Nero-Fuzzy inference system (ANFIS).were built to diagnose each pulse as normal or Ischemic, for comparing and choosing the best classifier.
The results of the system were 91.4% for sensitivity and 91.3% for accuracy for ANN classifier and 85% and 96% for ANFIS classifier respectively. When testing the proposed ANFIS system on some of chosen signals from the European Database, each one has duration of 10 minutes, the accuracy of detection was 90% of normal cases and 94% of ischemic cases.