Skip priors and add graph-based anatomical information, for point-based Couinaud segmentation

2025-08-03

AI Summary

This work addresses the reliance on manually annotated vascular anatomical priors in Couinaud liver segment segmentation. We propose a point-cloud-driven method that operates without explicit vascular topology input. Our key innovation is the first integration of a graph reasoning module into a point cloud segmentation framework, which implicitly learns hepatic vascular anatomical relationships by modeling point-wise neighborhood affinities—thereby enabling structural awareness of segmentation boundaries. The method jointly optimizes geometric and topological representations through an end-to-end architecture combining point cloud feature extraction with graph neural networks. Experiments on the MSD and LiTS datasets demonstrate that our approach achieves Dice scores and mean surface distances competitive with four state-of-the-art point-based methods, while significantly enhancing segmentation automation and preoperative planning accuracy.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Graph-based Machine Learning
• Reasoning under Uncertainty: Graphical Models

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Security and Privacy: Data transparency and provenance

Abstract

The preoperative planning of liver surgery relies on Couinaud segmentation from computed tomography (CT) images, to reduce the risk of bleeding and guide the resection procedure. Using 3D point-based representations, rather than voxelizing the CT volume, has the benefit of preserving the physical resolution of the CT. However, point-based representations need prior knowledge of the liver vessel structure, which is time consuming to acquire. Here, we propose a point-based method for Couinaud segmentation, without explicitly providing the prior liver vessel structure. To allow the model to learn this anatomical liver vessel structure, we add a graph reasoning module on top of the point features. This adds implicit anatomical information to the model, by learning affinities across point neighborhoods. Our method is competitive on the MSD and LiTS public datasets in Dice coefficient and average surface distance scores compared to four pioneering point-based methods. Our code is available at https://github.com/ZhangXiaotong015/GrPn.

Problem

• Eliminates need for prior liver vessel structure knowledge
• Improves Couinaud segmentation via graph-based anatomical learning
• Enhances point-based CT resolution without voxelization

Innovation

• Skip priors for Couinaud segmentation
• Add graph-based anatomical information
• Use point-based representations for CT