Segmentation of the hepatic arteries and veins

The vascularization of the liver is a useful information for the surgeon in order to localize accurately the lesions, especially metastases. Hepatic segmentation is based on the distribution of vessels according to the "Claude Couinaud" model. The objective of our work was to segment these hepatic vessels from Magnetic Resonance Imaging sections. First, we segmented the liver from the Magnetic Resonance Imaging slices using morphological operators via a nonlinear median filter and a "Disc" type structuring element. Then, two operations were performed: dynamic range cropping to improve the image contrast and histogram equalization to approach a flat histogram. Secondly, we segmented the hepatic vessels using the active contour model for the arteries segmentation and the level set algorithm for the veins segmentation.