An automated ECG classification system based on a neuro-fuzzy system

The 12-lead electrocardiogram (ECG), as well as the patient history, plays an important role in the early diagnosis of acute myocardial infarction (AMI). In this paper, a hybrid neuro-fuzzy approach to the diagnostic classification of 12-lead ECGs is presented. The architecture used is a combination of fuzzy logic and neural network theory. For ECG diagnosis, the system benefits from the reasoning capabilities of fuzzy logic as well as the learning ability of neural networks. This hybrid system consists of two phases: (1) Use fuzzy logic to establish the diagnosis system in the form of symbolic IF-THEN rules based on expert cardiac knowledge; (2) Through a training process, use a backpropagation network to automatically adjust the parameters of the system. A total of 124 ECGs from patients with or without acute myocardial infarction have been studied and eight diagnostic classes have been taken into account regarding the different locations of AMI. Sensitivity, specificity, partial and total accuracy are used for evaluation of the system. After the training process, the neuro-fuzzy system correctly identified 89.4% of the patients with AMI and 95.0% of the patients without AMI. The results confirmed that AMI can be diagnosed with reasonable accuracy. While we recognize that the diagnosis of AMI varies according to clinical circumstances, the hybrid system has the potential for automatic classification of AMI