Vascular tree segmentation in MRA images using Hessian-based multiscale filtering and local entropy thresholding
Ouazaa Hibet-Allah, Jlassi Hajer, Kamel Hamrouni · 2016
Segmentation of cerebrovascular structures from MRA (Magnetic Resonance Angiography) is a challenging assignment as a result of the complexity structures of the vessels and the characteristics of the MRA images. This article presents a new method which extracts the vascular structures from 2D medical images and helps the doctors in the treatment and diagnosis of vascular disease. In our proposed method, the Hessian-based multiscale filtering is firstly used to enhance vascular structures. And then, the blood vessels are extracted by executing the local entropy. Our method was tested on MRA database. Experimental results on MRA images demonstrate the ability to extract the most of the vascular structures successfully.