Analysis of myocardial infarction using wavelet transform and multiscale energy analysis

Myocardial Infarction (MI) is otherwise termed as heart attack, occurs when blood supply stops to certain artery or to some portion of arteries. MI is depicted in elevated ST-segment, wide pathological Q wave and inversion of T wave in electrocardiogram (ECG). This paper presents a multiscale energy based method for detection of MI. Detection of MI by consideration of fewer ECG leads requires prior information of the pathological characteristics of the disease. Thus, here we have considered all the 12 leads of the ECG signal simultaneously for detection of MI. Wavelet transform of all the leads of MI decomposes the signal into subbands of different order. The multiscale energy of all the bands are computed and the normalized multiscale energy of the wavelet coefficients are considered. The pathological structure present in the ECG data alters the covariance structure of the subband matrix and thus changes in the feature parameters of the signal occur, which leads to detection of MI. The results are presented using the standard MI ECG data from PTB diagnostic database.