Left ventricular wall deformation of CMR images using statistical analysis.

Abstract

Ventricular hypertrophy is one of the prevalent cardiovascular diseases. Quick and efficient diagnosis is the need of the hour and this paper attempts on an automated left ventricle wall deformation analysis which helps in the diagnosis of left ventricular hypertrophy. Intensity In-homogeneity Cardiac Magnetic Resonance (CMR) images are segmented by the proposed hybrid segmentation of Kirsch and modified Chan-Vese (CV) method with split-Bregman optimization. Co-occurrence features like contrast, correlation, energy and homogeneity are extracted from the segmented ventricular images. The features are then analysed in different directions of rotation for both normal and abnormal images in systolic as well as diastolic phases. Finally the area of the ventricular wall is obtained for both normal and abnormal images. The results thus obtained show that left ventricular wall deformation analysis using statistical features helps in the detection of pathological and physiological condition of the ventricles.