Our overall objective is to improve patient care by making quantitative cardiac measurement available in clinical practice. We have previously achieved substantial success in fast and accurate determination of left ventricle (LV) volume, ejection fraction and border location from echocardiograms. This technology will be modified and extended to determine regional wall thickness more accurately. The process uses a knowledgebase of the measured variation in geometry of normal and diseased human cardiac ventricles to rapidly and accurately reconstruct a patient's ventricle surface in three dimensions (3D) from echocardiogram images. In brief the approach is to capture 2D echocardiograms with known positions and orientations, to detect the ventricle surface borders on these images in 3D using in optimization program with a catalog of measured surfaces, and to compute the desired cardiac parameters from these border locations or from the computed 3D surface. The specific intent of this Phase I project is to demonstrate feasibility of improving wall thickness measurement by using a knowledge base of dual surface LV shapes. The accuracy of the measurements based on ultrasound will be verified with magnetic resonance imaging. In addition to accurate wall thicknesses, this approach to LV quantification will also provide regional shape information. This project will: Establish a test suite of 10 cases with geometry established from MRI. Analyze the ability of the dual surface catalog fit method to correctly estimate the geometry of each case.
