Design of a Form Factor Based Expert System for Detection of Myocardial Infarction Using Supervised Classification Technique

A computational approach for detection of myocardial infarction (MI) from ECG wave is presented. ECG signal of an infarcted person are different from that of healthy persons in terms of some specified spatial and temporal parameters. This results in difference in signal complexity for infarcted and healthy ECG. In this work a multiresolution wavelet based method is used for relevant features extraction and wave complexity is measured as form factor. A supervised pattern classification technique is applied for MI pattern recognition. The proposed expert system is tested and validated against 42 subjects taken from physionet database. The obtained sensitivity (Se), specificity (Sp) and accuracy are 95% and 90.8% and 92.8 respectively.