Delay Phase Hepatic Vessel Classify Methods In CT Images

Using medical image technology segments vessels from CT images and provide their 3D display, could help doctors observe vascular structures better.Hepatic vascular extracted from delay phase CT images. Because of the delay phase CT images contain a lot of noise naturally,the extracted vascular often appears adhesion and disconnection.To obtain more accurate results, vessels need to be classified into venous vein,portal vein and impurities.This paper researches a kind of classification method based on vessels segmented from delay phase CT images.Classification algorithm includes two parts,first to disconnect vessels adhesion part of the data,and then classify the vessels without adhesion.For vascular adhesion disconnect, this paper studied the disconnect method based on the center-line and disconnect method based on region.Method based on the center-line uses the special angle relationship of adhesion center-line to complete disconnect.Method based on region uses regional local mean value, variance and gradient to remove adhesion points,obtain no adhesion vessels by restore the separated vessels.To classify the vessels without adhesion,this paper realizes a kind of classification method based on hepatic vessels topology feature.First using the features portal vein trunk located outside of hepatic and has high roundness to identify the portal vein,then classify the hepatic venous,impurities and small portal vein branches by branch number changes and trend.Apply our method to many different vessels segmented from delay phase CT images.The experimental results show that the method can realize automatic classification of hepatic vessels ideally.