Accurate Vessel Segmentation with Progressive Contrast Enhancement and Canny Refinement

Vessel segmentation is a key step for various medical applications, such as diagnosis assistance, quantification of vascular pathology, and treatment planning. This paper describes an automatic vessel segmentation framework which can achieve highly accurate segmentation even in regions of low contrast and signal-to-noise-ratios (SNRs) and at vessel boundaries with disturbance induced by adjacent non-vessel pixels. There are two key contributions of our framework. The first is a progressive contrast enhancement method which adaptively improves contrast of challenging pixels that were otherwise indistinguishable, and suppresses noises by weighting pixels according to their likelihood to be vessel pixels. The second contribution is a method called canny refinement which is based on a canny edge detection algorithm to effectively re-move false positives around boundaries of vessels. Experimental results on a public retinal dataset and our clinical cerebral data demonstrate that our approach outperforms state-of-the-art methods including the vesselness based method [1] and the optimally oriented flux (OOF) based method [2].