Vascular stenosis detection based on morphological features

Vascular stenosis detection is important for the prevention of ischemia and necrosis of tissues and organs. In this paper, we proposed a novel method to detect the vascular stenosis which is based on morphological features. First, Six morphological methods, based on boundary and area respectively, are proposed. Then these morphological methods are used to extract the features of each vascular cross-section. Finally, the decision tree classifier is used to analyze and judge the vascular stenosis. The experimental results on CTA images indicate that the proposed method is more sensitivity and objectivity than current automatic methods. It can solve the problem of difference in vascular stenosis detection. The accurate rate can achieve 87% in 230 vessels. Using this method, doctors can save lots of time and also the misdiagnosis can be reduced.