Accurate Pelvis and Femur Segmentation in Hip CT With a Novel Patch-Based Refinement

Area of Science:

• Medical Imaging
• Computer-Aided Diagnosis
• Orthopedic Surgery

Background:

• Accurate segmentation of pelvis and femur in hip computed tomography (CT) images is challenging due to bone deformation and joint space narrowing in diseased hips.
• Existing methods struggle with the complexities of pathological hip anatomy.

Purpose of the Study:

• To develop a fully automatic segmentation framework for pelvis and femur in both healthy and diseased hip CT images.
• To enhance segmentation reliability for diseased hips, particularly those with significant bone deformation and joint space narrowing.

Main Methods:

• A three-step framework: preprocessing to extract the volume of interest (VOI), coarse bone segmentation using a conditional random field (CRF) model, and a novel patch-based refinement method.
• The refinement method utilizes neighborhood information for label fusion, improving accuracy and robustness.
• The approach was validated on 60 CT datasets (78 healthy, 42 diseased hemi-hips) using three-fold cross-validation.

Main Results:

• The proposed framework achieved significantly increased segmentation accuracy for diseased hemi-hips compared to two state-of-the-art methods.
• The novel patch-based refinement method demonstrated particular suitability for diseased hips with bone deformation.
• The method proved more accurate and robust for diseased hips exhibiting joint space narrowing.

Conclusions:

• The developed automatic segmentation framework is highly effective for both healthy and diseased hip CT images.
• This method offers a significant improvement for the automatic segmentation of diseased hips, addressing limitations of current techniques.
• The framework shows promise for clinical applications requiring precise hip joint segmentation.