This research will determine the feasibility of automatically detecting the borders of the left ventricle (LV) from a three-dimensional (3D) echocardiographic scan. The goal is to delineate the endocardium and epicardium as 3D surfaces for use in measuring cardiac parameters. The approach is to fit a 3D model of the LV and associated structures to the image data. The hypothesis is that incorporating knowledge of the range of heart shapes observed among normal subjects and patients with diseased hearts into a 3D model will facilitate automated border detection (ABD) by constraining the process to only yield heart-like shapes. The investigators have already developed the computer algorithms for reconstructing the LV endocardial and epicardial surfaces in 3D and for calculating the mean and covariance matrix for the 3D shape of the LV based on data acquired from a sample of 3D reconstructions. Additional 3D scans of 51 normal subjects and 35 patients with diseased hearts have been acquired and partially analyzed.
The specific aims of the proposed work are: 1) to complete the ultrasound model of the gray scale appearance of the heart in any plane: the right ventricle, pericardium, and papillary muscles will be added to the current LV model; 2) to complete the algorithms for feature extraction from images, and for mesh optimization the procedure using the extracted features to fit the 3D model to the patients image data; and 3) to demonstrate the feasibility of the ABD process using simulated images. In Phase II these components of the 3D ultrasound border detection process will be tested and modified as necessary to yield clinical acceptable levels of accuracy and performance. In Phase III a quantitative 3D echocardiographic system will be developed by installing the ABD algorithm into a hardware and software system for acquiring and analyzing 3D echo scans of the heart. This research is important because the improved accuracy afforded by 3D echo cannot be applied to clinical practice without automation.
The
for this grant is to develop a hardware and software system for acquiring three dimensional echocardiograms and for performing quantitative analysis of the size, shape, and function of the left ventricle.
