Automatic Intrahepatic Biliary Segmentation Based Image Processing

Biliary tree is a set of tubes that connects the liver to an organ below it directly that is called Gallbladder. Biliary tree contains a main tube known as bile duct. The dilation of a bile duct is an important indicator for other serious issues in the human body such as stones and tumors, which commonly occur due to pancreas or papilla of vater. In many cases, the discovery of dilation in bile duct is difficult to be recognized by novice or unspecialized medical staff. Even specialists cannot recognize the dilation of bile duct with naked eye.  This paper provides a novel vision-based framework to help in initial diagnosis of the biliary tree. The framework applied several image processing techniques in order to segment the biliary tree from the Magnetic Resonance Image (MRI). Once the region of interest segmented from the image, several computations applied on it to extract five features including major axis, minor axis, bile duct area, biliary tree area and compactness. A database of images from King Hussein Medical Center (Amman-Jordan) is used in this research, containing 200 MRI images; 100 normal cases; and 100 cases with dilated bile ducts. After the features are extracted, multiple classifiers are applied to obtain the status of the patients in term of (normal or dilated). The results show that the extracted features work well with all classifiers with large accuracy and area under curve values. The novelties of this work include introducing an automated process to segment the biliary tree from MRI images, in addition to scientifically correlate extracted features that have not previously introduced in the literature with the status of biliary tree.