Semi-automated visualization of coronary arteries from three-dimensional magnetic resonance images

摘要

Visualization of coronary arteries from three-dimensional (3D) magnetic resonance (MR) coronary angiography often requires manual delineation of the vessels, a tedious and time consuming task. In this work, a semi-automated technique to rapidly identify the path of one or more coronary arteries throughout a 3D volume is presented. The method applied a nonlinear morphological filter to enhance vessel voxels and suppress myocardial muscle and blood pool. The filtered data and a user-selected ROI were then used to formulate a directed graph, which was subsequently searched to find an optimum path for the vessel through the volume. The coronary vessel was viewed by a curved reformatting and maximum intensity projection (MIP) along the derived path. The algorithm was applied to 3D MR volumes of left and right coronary arteries and successfully produced complete visualizations of the vessels of interest, demonstrating that morphological filtering and optimal path searching were capable of efficiently extracting and viewing coronary vessel paths from large 3D volumes.