17 February 2012 Automatic contour and centerline extractions of single and bifurcated vessels in coronary angiogram
Author Affiliations +
Abstract
We propose an automatic contour and centerline extraction methods of single and bifurcated vessels in coronary angiograms. Our method consists of four steps. First, for enhancing vascular structures, anisotropic diffusing filtering and Hessian-based multi-scale filtering are performed. Second, an initial vessel region is segmented by region growing with adaptively defined threshold value. Third, the initial vessel region is thinned and pruned for extracting an initial vessel centerline. For bifurcated vessel analysis, a bifurcated polygon is defined in a bifurcated lesion and a bifurcated point is detected by matching of the bifurcation patterns from vessel centerline. Finally, contrast is stretched in the ROI except lower 10 and upper 30 percent density ranges. Then vessel contour points are extracted by canny edge detector and are selected by the crossing contour points perpendicular to the vessel centerline. In the inner contour with high curvature and weak contrast of the bifurcated polygon, the vessel contour points are selected by the crossing contour points radial to the vessel centerline. Experimental results show that our method provides accurate results in narrow sections such as occlusion or stenosed vessels and ignores overlapped regions such as diaphragm and crossing vessels. For bifurcated area, the bifurcated point and vessels are well-extracted in low-contrast and non-uniform illumination.
© (2012) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Junga Baek, Helen Hong, "Automatic contour and centerline extractions of single and bifurcated vessels in coronary angiogram", Proc. SPIE 8316, Medical Imaging 2012: Image-Guided Procedures, Robotic Interventions, and Modeling, 83162T (17 February 2012); doi: 10.1117/12.911772; https://doi.org/10.1117/12.911772
PROCEEDINGS
8 PAGES


SHARE
Back to Top