Optimized Coronary Artery Segmentation Using Frangi Filter and Anisotropic Diffusion Filtering

Abstract

X-ray angiography is currently the prime method of diagnosis during percutaneous coronary interventions. Robust automatic detection of coronary arteries from angiography images is of great interest. Hessian-based Vessel enhancement filtering was proven successful in automatically segmenting vessels from angiography images. However, there is too much noise and other anatomical structures also interfere with the Percutaneous Coronary Intervention (PCI) procedure. The proposed method uses Frangi Hessian based vessel enhancement filter for extracting coronary arteries and setting optimal value of Frangi filter parameters a and ß. The method is applied recursively on a set of angiography images from same machine and tries to assign optimal values of a and ß for it. This procedure is followed by (Rotation Invariant) An isotropic diffusion filtering of the image. An isotropic Diffusion Filtering is used for noise removal and coronary artery enhancement. For the diffusion tensor, hybrid diffusion is used with a continuous switch which is suitable for filtering tubular image structures.