Discrimination of different myocardial infarction stages using wide band electrocardiogram

Abstract

Abstract Most of usual electrocardiogram (ECG) signals normally keep the signal energy in the 0.05–100 Hz band, but higher frequencies containing valuable diagnostic information are also present in wide band (WB) ECG signals. Existing studies on computer-assisted myocardial infarction (MI) diagnosis are mostly based on the usual ECG signals, and the valuable diagnostic information has not been used sufficiently yet. Multivariable autoregressive coefficients were extracted from WB orthogonal ECG (OECG) signals for the classification purpose in this research. The data for the analysis were taken from Physikalisch Technische Bundesanstalt diagnostic ECG database including health control, MI in early stage and MI in acute stage. In order to further investigate the performance of WB ECG signals, standard ECG signals with a wide frequency band were utilized for the feature extraction and classification as the same way. The experimental results showed that the MI classification accuracy could be improved by introducing WB ECG signals and the features extracted from WB OECG signals with a frequency of 0–250 Hz would be the best efficient representation for discriminating different MI stages. The classifiable and efficient features can be extracted from WB OECG signals for the classification of MI stages with a high accuracy.