Automatic multi-parametric quantification of the proximal femur with quantitative computed tomography

Area of Science:

• Orthopedics and Bone Health
• Medical Imaging and Radiology
• Computational Anatomy

Background:

• Quantitative computed tomography (QCT) is crucial for assessing proximal femur bone quality.
• Key assessments include volumetric bone mineral density (vBMD), tissue volume, bone strength (finite element modeling - FEM), and cortical bone thickness.

Purpose of the Study:

• To present an automated framework for multi-parametric QCT quantification of the proximal femur.
• To enable accurate and reproducible computational anatomy studies of bone quality.

Main Methods:

• An automated framework was developed for proximal femur QCT analysis.
• This involved scan cropping, multi-atlas segmentation, template registration, and parametric mapping (vBMD, tissue volume, FEM strength, cortical thickness).

Main Results:

• High accuracy was achieved with a Dice similarity coefficient of 0.976±0.006.
• Excellent reproducibility was demonstrated with CVRMS values generally below 4% for various QCT parameters across different sites and manufacturers.

Conclusions:

• The automated QCT framework provides accurate and reproducible multi-parametric analysis of the proximal femur.
• This approach is highly valuable for multi-site studies, including clinical trials involving elderly populations.