Automatic Multi-Atlas Liver Segmentation and Couinaud Classification from CT Volumes

Area of Science:

• Medical Imaging
• Computational Anatomy
• Oncology

Background:

• Primary Liver Cancer (PLC) is the sixth most common cancer globally, often associated with chronic liver diseases and viral hepatitis.
• Effective treatment for PLC and malignant liver tumors necessitates individualized plans based on tumor characteristics and liver function.
• Accurate liver segmentation and classification (e.g., Couinaud's classification) are critical for computer-aided diagnosis and treatment planning.

Purpose of the Study:

• To develop and validate an automatic liver segmentation and Couinaud classification method using computed tomography (CT) imaging.
• To overcome the time-consuming and user-dependent limitations of manual liver segmentation.
• To enhance accuracy and efficiency in analyzing liver anatomy for clinical applications.

Main Methods:

• Implementation of a multi-atlas based classification approach for automated liver segmentation.
• Application of the algorithm to a dataset of 20 subjects from the IRCAD 3D database.
• Validation using the DICE coefficient to measure segmentation accuracy.

Main Results:

• The proposed automatic method achieved an average DICE coefficient of 0.94 for liver segmentation and classification.
• Demonstrated significant improvements in accuracy and time consumption compared to manual methods.
• Successfully segmented and classified liver volumes according to Couinaud's classification.

Conclusions:

• The developed automatic multi-atlas liver segmentation and Couinaud classification method is accurate and efficient.
• This technique offers a valuable tool for computer-aided diagnosis and personalized treatment planning in Primary Liver Cancer (PLC) patients.
• CT image analysis combined with this automated approach can streamline liver volume assessment.