Myocardial infarction detection using hybrid BSS method

This paper presents a novel technique for detecting myocardial infarction (MI) from ECG. Today, MI is one of the major causes of death worldwide. MI is a minor stage in lifelong cardiac disease, although rarely noticed, but can lead to sudden death. Since it is the first symptom of coronary artery disease, it is very important to detect MI in an early stage. For this purpose, several methods have been employed. But the hurdle is to detect MI efficiently from noisy signal, and also in an early stage. The problem lies in data acquisition system. Several noises get mixed with ECG signal, at the time of data acquisition and this make the extraction difficult. These noises can be due to respiration, EMG etc. The inefficient low pass filters of ECG machine add these noises to ECG signal. So to get a good input data, a perfect denoising technique is required, which is inevitable for the detection of small variations. The existing techniques use filters like notch and adaptive for denoising since they are not considering this raw ECG. It is proved that Independent Component Analysis (ICA) based on blind source seperation (BSS) algorithms, is an effective tool for extracting ECG from a mixed raw signal. Better denoising gives better results for the early detection of MI. Thus the objective of this work is to detect myocardial infarction in an early stage. The classification is based on naïve bayes classifier. The presented algorithm gives 96.77% accuracy on PhysioNet ECG database, which is supported by the National Institute of General Medical Sciences (NIGMS) and the National Institute of Biomedical Imaging and Bioengineering (NIBIB).