Graph cut liver segmentation for interstitial ultrasound therapy

Area of Science:

• Medical Imaging
• Computational Anatomy
• Oncology

Background:

• Percutaneous high intensity ultrasound surgery is a primary curative treatment for liver cancer.
• Accurate segmentation of liver, tumor, and vasculature is crucial for treatment planning.
• Existing segmentation methods can be time-consuming and labor-intensive.

Purpose of the Study:

• To propose a fast 3D semi-automatic segmentation method for liver, tumor, and hepatic vascular networks.
• To improve the efficiency and accuracy of pre-surgical planning for liver cancer treatment.
• To leverage graph-based image analysis for medical volume segmentation.

Main Methods:

• A graph description of contrast-enhanced CT volumes was utilized.
• Links between nodes represented voxel similarity (region-based) or class changes (boundary-based).
• A Max-Flow/Min-Cut graph cut algorithm partitioned the volume after interactive training.

Main Results:

• The proposed method enables fast and semi-automatic segmentation of key anatomical structures.
• The graph-based approach effectively captures both regional and boundary information.
• Successful partitioning of CT volumes into segmented classes was achieved.

Conclusions:

• The developed 3D semi-automatic segmentation method is efficient for liver cancer treatment planning.
• This technique aids in precise delineation of the liver, tumor, and vasculature for percutaneous high intensity ultrasound surgery.
• Graph cut algorithms offer a powerful tool for complex medical image segmentation tasks.