Modeling left ventricle shape from 2D CT images using wavelets and mean regular hexagon

Abstract

We present modeling of left ventricle (LV) shape from 2D CT images, acquired from a 64-slice CT medical imaging modality, using wavelets, regular hexagonal boundary tracing, and simple statistical moments. Our proposed algorithm uses a regular hexagonal approximation model corresponding to interactive labeling of LV segmented texture using wavelet-based techniques. We generate LV shape model, as the mean regular hexagonal approximation. We demonstrate application and usefulness of our shape model for successful identification of end-diastolic and end-systolic stages of a normal human cardiac cycle.