Analysis of myocardial infarction in CMR images using hybrid level set based segmentation and regional ventricle contractility analysis

Abstract

The assessment of left ventricle (LV) wall motion plays a major role in the diagnosis of myocardial infarction (MI). The aim of this work is to study regional contractility of LV in MI and normal subjects using magnetic resonance images. The segmentation of ventricular cavity is performed with corr-entropy based local bias field corrected image fitting (CELBIF) method. Myocardial contraction over a cardiac cycle is estimated for each sector based on Hausdorff distance and wall motion score index. The results show that CELBIF algorithm yields higher value for Dice coefficient (0.92) than LBIF method. The tracking of LV shows an increase in ventricular volume in infarcted subjects for entire cardiac cycle. Lower contraction is observed in infarcted LV cavities due to damage in myocardium sectors. The ventricular tracking and clinical indices detect abnormal cardiac behavior in MI subjects. The regional contractility analysis aids the identification of infarcted myocardial segment.