Noninvasive localization of myocardial infarction by means of a 3D heart-model based inverse imaging approach

It is of significance to localize sites of myocardial infarction (MI) noninvasively. The authors have investigated the feasibility of localizing MI sites from body surface ECGs in terms of a heart-model based new inverse solution. A preliminary diagnosis system was developed based on an artificial neural network (ANN) to limit the searching space of the optimization algorithm and to set the initial parameters in the heart simulation model. The optimal heart-model parameters were obtained by minimizing the difference between the observed and model-generated body surface ECGs. The present computer simulation shows promising results indicating that the present approach produces a 100% success rate in localization of an area of acute MI in the left ventricle for cases studied. The present study suggests this approach merits further investigation, and may provide an important alternative to ECG inverse solutions.