An automated evaluation of regional left ventricular function on cine magnetic resonance images

Abstract

Routine evaluation of the regional ventricular function in cardiac magnetic resonance (CMR) remains mostly based on visual analysis of cine loops and subsequent interobserver variability remains a potential drawback for diagnosis and follow-up of coronary artery disease (CAD). In this paper we propose a method that ensures automated assessment of normalized mean contraction times and radial endocardial velocities from MR cine images, by combining a parametric imaging analysis to an automatic segmentation method of the left ventricle. Short-axis cine loops were obtained in 20 subjects (10 myocardial infarction (MI) and 10 control). Reference segmental wall motion was scored by an expert according to 4 classes (normokinesia, hypokinesia, akinesia and dyskinesia). Results showed increased values for normalized mean contraction times and decreased values for the radial velocity in segments with wall motion abnormalities (WMA) compared to segments with normal function. These functional parameters could help to detect regional WMA.