A Research On Bile Duct Segmentation In Abdominal CT Images

In clinical practice,doctors diagnose biliary system diseases and plan treatment by reading abdominal CT images,where biliary system images are required to be set apart.Therefore,it will be vitally helpful for doctors to research on segmentation methods of biliary system and develop an according system.Difficulties of bile duct segmentation lies the spatial and CT value proximity between bile duct and ligament tissues,as a result of which they are easily confused.making use of ligaments' mainly being located at the hepatic portal,surface structure,and the slightly higher CT value than that of bile duct there and the difference between the CT value of intrahepatic bile duct and hepatic portal one segmentation method by processing based on subarea could be put forward,which is segmenting intrahepatic bile duct and hilar bile duct respectively and adjoining them together.Concrete procedures including four parts,generating images of liver mask and hepatic portal mask,segmenting and adjoining images of bile duct and hepatic duct and processing after segmentation.When generating intrahepatic mask images,based on the existing liver segmentation results i.e.original liver mask,Boundary tracking and scanning line method could be adopted to fill the holes in the original liver mask quickly.By calculating the convex shell of mask,the precipitously recessed area to be filled could be determined,thus hepatic portal mask obtained.Under the liver mask,adopt the bile duct segmentation result through threshold segmentation.For the hepatic portal,segment with the threshold in accordance to the CT value of common bile duct and remove ligaments with thin surface through opening operation.Given that opening operation would remove small bile duct,this essay proposes a strategy of growing from bile duct to common bile duct through controlling incidence of growth and tubular constraint,hence ensuring that the growth would not extend to the ligaments and repairing the disconnection between bile duct and common bile duct.Finally,measures of small connection region removal,surface structural region deletion and liver surface targets removal would be implemented to the bile duct segmentation results.The ultimate result would be obtained by limiting the spatial area of growth.Tests have been carried out with the entire experimental data provided by cooperating partner of the project.The result shows that the system could segment the bile duct expeditiously and precisely,which is recognized by the cooperating partner.