A Review on Ventricular Segmentation for Cardiac MR Images

Abstract

Cardiac MRI is most comprising process for both functionality estimation and anatomized structure of the heart. The delineation of various ventricles are obtained from experienced physicians.  Therefore an efficient method being invented which can replace human observer variability. Fully automatic delineation algorithms are on high demand. Sectionalization  of Left and Right Ventricles is imperative part in quantitative analysis in CMRI.  This work presents study of various techniques used for right, left and biventricular delineation with performance evaluation parameters such as Dice Metric, Haussdorff Distance and Ejection Fraction. The result of LV delineation is quite solved as compared to RV and Bi-ventricle segmentation. The irregular shape of RV affects the accuracy of models.  The robustness and efficiency of various algorithms is challenging due to lack of publically available huge data sets for healthy and pathological patients.