Fully automated quantification of left and right ventricular volumes throughout the cardiac cycle from magnetic resonance imaging

The left ventricle (LV) has been widely studied and many segmentation methods have been proposed allowing accurate volume estimation from cardiac magnetic resonance imaging (CMRI). Anatomical complexity of the right ventricle (RV) has made accurate determination of RV shape and volume difficult. We propose a fully automated method for both RV and LV segmentation from CMRI. Datasets were analysed using custom software based on a region-based level set model followed by a regularization motion, allowing dynamic endocardial contour detection. RV and LV contours were also manually traced at end-diastole (ED) and end-systole (ES). For both techniques, disk-area summation method was applied to compute volumes. Volumes at ED and ES and ejection fraction were compared. The results of this initial study provide the basis for automated, fast and accurate quantification of LV and RV size, function and volume change throughout the cardiac cycle.