Multiple feature-based portal vein classification for liver segment extraction.

Area of Science:

• Medical Imaging
• Computational Anatomy
• Surgical Planning

Background:

• The Couinaud classification method relies on portal vein (PV) third-order branches to define liver segments, crucial for hepatic resection surgery.
• Accurate reconstruction and classification of PV are challenging due to complex vascular structures, impacting surgical planning.
• Existing methods face difficulties in precisely segmenting and classifying the intricate PV network.

Purpose of the Study:

• To develop an automated method for accurate separation and classification of portal vein branches.
• To enable precise extraction of liver segments based on the Couinaud classification system.
• To improve the functional anatomical understanding of the liver for pre-operative planning.

Main Methods:

• A multiple feature-based approach is proposed for liver segment extraction.
• Complete PV separation is achieved using minimal node cut strategies with fused topology and appearance features.
• Bifurcation nodes are detected, ordered, and refined using vascular topology to classify branches and identify third-order PV branches.

Main Results:

• The method demonstrated high accuracy on CT and MR datasets, with average Dice, Jaccard, Recall, and Precision values of 93.00%, 87.90%, 93.47%, and 93.19%, respectively.
• Separation results are more accurate compared to state-of-the-art methods.
• Eight liver segments corresponding to distinct functional areas were precisely extracted based on classified third-order PV branches.

Conclusions:

• The proposed method achieves high accuracy in liver segment extraction.
• Precise liver segment extraction is significant for pre-operative planning in hepatic resection surgery.
• This technique offers a reliable tool for understanding liver functional anatomy.