A multiatlas segmentation using graph cuts with applications to liver segmentation in CT scans

Area of Science:

• Medical Imaging
• Computer-Aided Diagnosis
• Image Segmentation

Background:

• Accurate segmentation of organs like the liver is crucial for medical diagnosis and treatment planning.
• Existing atlas-based segmentation methods can be limited by registration accuracy and atlas selection strategies.

Purpose of the Study:

• To develop and evaluate a novel atlas-based segmentation approach for liver segmentation in computed tomography (CT) images.
• To improve segmentation accuracy by combining probabilistic and multiatlas segmentation techniques.

Main Methods:

• A hybrid approach integrating low-level operations, an affine probabilistic atlas, and multiatlas segmentation.
• Spatial normalization and energy function minimization using graph cuts for segmentation.
• Region of Interest (ROI) definition and coarse segmentation guiding multiatlas registration and selection.

Main Results:

• The proposed method achieved highly accurate liver segmentation in CT images.
• Performance surpassed traditional atlas selection and nonrigid registration methods within the defined ROI.
• Segmentation results were comparable to those of human experts and recent state-of-the-art methods.

Conclusions:

• The combined probabilistic and multiatlas segmentation approach offers a robust and accurate solution for liver segmentation.
• The method's reliance on coarse segmentation for guiding registration and selection enhances overall performance.
• This technique holds promise for improving computer-aided diagnosis in medical imaging.