Computational Anatomy Atlas of the Heart

Abstract

A computational anatomy atlas of the heart has been built for the main purposes of patient-specific functional analysis and electromechanics simulations. The atlas is composed of an underlying statistical shape model, based on a point distribution model (PDM), which was constructed from a training set comprising 100 asymptomatic and pathologic subjects. A key advantage of the proposed building procedure is the needless of manual delineations. Another one is the use of multi-slice computed tomography (MSCT) images, which have an excellent spatial resolution. Both, inter-subject and temporal variability were captured with the inclusion of all available time frames of each 4-D (3-D + time) image. Embedded in the shape model, a number of functional structures and properties have been incorporated to the atlas, such as myocardial fiber orientation, specialized electrical conduction pathways in the ventricles and atria, a set of anatomical landmarks, and the coronary venous and arterial trees. While numerous potential improvements are planed for the future, the atlas is already workable for FEM-based simulations of ventricular electrophysiology and cardiac functional analysis.