Accurate Coronary Vessel Extraction Using Fast Directional Filter Bank

Correct discern of blood vessels performs a significant role in a number of clinical applications, especially in angiographic image analysis. In this paper, we propose a new simple and fast fabrication of DDFB (called FDFB) without necessity of downsampling and resampling employed in DDFB, which is time consuming, especially if the directions are sizeable. Also, in this paper we discuss on parameters of vessel filter in more detail and propose a proper range for parameters selection to achieve accurate vessel enhancement. One major problem in automatic segmentation is the presence of background non-vessel structures; we use a morphological-based algorithm to remove these, after vessel enhancement. Another problem with DDFB is the ringing artefact; smoothed FDFB unravels the later. The experimental results show that the proposed FDFB and vessel enhancement framework work well on angiograms.