Hip-Joint CT Image Segmentation Based on Hidden Markov Model with Gauss Regression Constraints

Area of Science:

• Medical Imaging
• Computer Vision
• Biomedical Engineering

Background:

• Hip-joint CT images suffer from low contrast, irregular boundaries, and noise.
• Traditional segmentation methods are inefficient and require manual input, limiting clinical utility.
• Existing fuzzy clustering algorithms are sensitive to image noise.

Purpose of the Study:

• To develop an efficient and accurate hip-joint CT image segmentation algorithm.
• To overcome the limitations of traditional methods and improve segmentation accuracy.
• To reduce manual intervention in medical image analysis.

Main Methods:

• Proposed a fuzzy clustering algorithm integrating Gaussian regression model (GRM) and hidden Markov random field (HMRF).
• Utilized prior information to regularize the fuzzy C-means objective function, enhanced with KL information.
• HMRF established label field neighborhood relationships; GRM established feature field neighborhood relationships based on pixel label consistency.

Main Results:

• The proposed algorithm demonstrated high segmentation accuracy for hip-joint CT images.
• The combined GRM-HMRF approach effectively handled low contrast and image noise.
• Reduced sensitivity to noise compared to classical fuzzy clustering methods.

Conclusions:

• The novel fuzzy clustering algorithm offers a significant improvement for hip-joint CT image segmentation.
• The integration of GRM and HMRF provides a robust solution for noisy medical images.
• The method shows potential for enhanced clinical applications requiring precise image segmentation.