CTA CORONARY LABELING THROUGH EFFICIENT GEODESICS BETWEEN TREES USING ANATOMY PRIORS
Embargo until
Date
2014-10-23
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Johns Hopkins University
Abstract
We present an efficient realization of recent work on unique geodesic paths between tree shapes for the application of matching coronary arteries to a standard model of coronary anatomy in order to label coronary centerlines extracted in cardiac Computed Tomography (CT) Angiography (CTA) data. Automatically labeled coronary arteries would speed reporting coronary diseases for physicians, be used for building patient specific myocardial segment models for the correct integration of coronary anatomy with myocardial function and guide segmentation algorithms for extracting the centerline representation of coronary arteries in CTA data.
Our approach builds on Quotient Euclidean Distance metric that leverages both geometric and topological information in order to compute unique natural and continuous geodesic deformations between tree-shapes. The efficiency of our approach and the quality of the results are enhanced using the relative position of detected cardiac structures including four chambers and pericardium. We explain how to efficiently compute the geodesic paths between tree shapes using Dijkstra's algorithm and we present a methodology to account for missing side branches during matching. We address computational difficulties for labeling large and complex coronary arteries and present evaluation results on 50 expert annotated and 20 automatically detected coronary centerlines in CTA data. For nearly all labels our approach shows promise compared with recent work and we show results for 12 additional labels. The results show the practicality and accuracy of our approach for labeling patient specific coronary centerlines extracted in CTA.
Advisors:
Dr. Russell H. Taylor from JHU
Dr. Gareth Funka-Lea from Siemens
Description
Keywords
coronary labeling, shape space, tree matching