Automated CT segmentation of diseased hip using hierarchical and conditional statistical shape models

Area of Science:

• Medical imaging and biomechanics
• Orthopedic surgery and computational anatomy

Background:

• Accurate segmentation of the femur and pelvis is essential for patient-specific surgical planning and simulation in hip procedures.
• Diseased hip joints present challenges due to deformed shapes and narrow joint spaces, complicating boundary determination of the femoral head and acetabulum.

Purpose of the Study:

• To develop and validate a multi-stage segmentation method for improved accuracy in diseased hip joints.
• To enhance the segmentation of the femoral head and acetabulum, critical components for hip surgery.

Main Methods:

• A hierarchical hip statistical shape model (SSM) was employed for initial segmentation of the pelvis and distal femur.
• A conditional femoral head SSM was subsequently utilized, leveraging previously segmented regions for improved accuracy.
• The method was validated using CT data from 100 diseased patients (200 hemi-hips) across various disease types and severities.

Main Results:

• The multi-stage method demonstrated significantly increased segmentation accuracy for the femoral head.
• The approach effectively addressed challenges posed by deformed shapes and narrow joint spaces in diseased hips.

Conclusions:

• The proposed multi-stage SSM approach offers a robust solution for accurate hip joint segmentation in clinical settings.
• This method has the potential to improve the precision of patient-specific planning and simulation for hip surgeries.