FACTS: Fully Automatic CT Segmentation of a Hip Joint

Area of Science:

• Medical imaging
• Computer-assisted surgery
• Orthopedic surgery

Background:

• Accurate hip joint surface models from CT data are crucial for computer-assisted diagnosis and planning (CADP) of periacetabular osteotomy (PAO).
• Current CADP systems often rely on manual segmentation, which is time-consuming and lacks reproducibility.

Purpose of the Study:

• To present a Fully Automatic CT Segmentation (FACTS) approach for simultaneous extraction of pelvic and femoral models.
• To enable fully automatic initialization of multi-atlas segmentation and preserve hip joint structure.

Main Methods:

• Combined fast random forest (RF) regression-based landmark detection using improved fast Gaussian transform (IFGT).
• Multi-atlas based segmentation initialized automatically.
• Articulated statistical shape model (aSSM) based fitting to prevent model penetration.

Main Results:

• Mean segmentation accuracy of 0.40 mm (pelvis), 0.36 mm (left femur), and 0.36 mm (right femur) compared to manual segmentation.
• Differences in PAO parameters: 2.0° ± 1.5° (anteversion), 2.1° ± 1.6° (inclination), and 3.5% ± 2.3% (acetabular coverage).
• Results are clinically accurate for target applications.

Conclusions:

• The FACTS approach offers a fully automatic and reproducible method for hip joint segmentation.
• The method achieves high accuracy, suitable for clinical application in PAO diagnosis and planning.
• This automated approach can significantly improve efficiency and reliability in orthopedic surgery planning.