Automatic liver segmentation on Computed Tomography using random walkers for treatment planning.

Area of Science:

• Medical Imaging
• Computational Anatomy
• Radiology

Background:

• Liver segmentation from Computed Tomography (CT) is vital for liver disease treatment planning.
• Accurate segmentation is challenging due to indistinct liver boundaries, variable intensity patterns, and anatomical differences, especially in pathological livers.

Purpose of the Study:

• To develop an accurate and efficient framework for segmenting contrast-enhanced liver CT images.
• To automate liver segmentation, reducing reliance on manual initialization and improving accuracy for diverse liver conditions.

Main Methods:

• A random walker-based framework was employed for liver segmentation.
• Automatic detection of the liver dome using the right lung lobe's location.
• Computational efficiency was enhanced through rib-caged area segmentation prior to liver extraction.

Main Results:

• The proposed method achieved high accuracy on a mixed dataset of healthy and pathological livers (Overlap Error: 4.47%, Dice Similarity Coefficient: 0.94).
• Exceptional accuracy was demonstrated for pathological livers (Overlap Error: 5.95%, Dice Similarity Coefficient: 0.91).
• The method offers competitive performance compared to existing liver segmentation techniques.

Conclusions:

• The random walker framework provides an accurate and automated solution for liver segmentation in CT images.
• This approach effectively addresses the challenges posed by pathological liver variations.
• The automated detection and segmentation improve efficiency and accuracy in clinical applications.