Automatic liver vessel segmentation using 3D region growing and hybrid active contour model

Area of Science:

• Medical Imaging
• Image Processing
• Computational Anatomy

Background:

• Accurate liver vessel segmentation is crucial for diagnosing and treating liver diseases.
• Existing 3D segmentation methods often struggle with complex vessel structures and thin vessel details.

Purpose of the Study:

• To develop and validate a novel, automated 3D method for liver vessel segmentation.
• To improve the accuracy and completeness of thin vessel segmentation in abdominal CTA images.

Main Methods:

• A hybrid approach combining 3D region growing with a bi-Gaussian filter for thin vessels.
• Integration of a hybrid active contour model and K-means clustering for thick vessel segmentation.
• Validation using abdominal computed tomography angiography (CTA) datasets.

Main Results:

• Achieved high performance metrics: 98.2% accuracy, 68.3% sensitivity, 99.2% specificity, 73.0% Dice, 66.1% Jaccard, and 2.56 mm RMSD.
• Demonstrated superior segmentation of complex liver vessels, preserving continuous and complete thin vessel structures.
• Outperformed several existing 3D vessel segmentation algorithms in experimental evaluations.

Conclusions:

• The proposed automated method effectively segments 3D liver vessels with enhanced detail for thin structures.
• This technique offers a robust and accurate solution for liver vessel segmentation in medical imaging.
• The method shows significant potential for clinical applications in liver disease assessment.