Automated cortical bone segmentation for multirow-detector CT imaging with validation and application to human...

Area of Science:

• Biomedical Engineering
• Radiology
• Orthopedics

Background:

• Cortical bone is crucial for skeletal function, bone strength, and fracture risk assessment.
• In vivo assessment of cortical bone using peripheral multirow-detector CT (MD-CT) imaging is valuable but faces challenges like low resolution and complex bone structures.
• Accurate segmentation of cortical bone is essential for quantitative analysis.

Purpose of the Study:

• To develop and evaluate a fully automated algorithm for segmenting cortical bone at the distal tibia from in vivo MD-CT imaging.
• To assess the performance and applicability of the developed algorithm.

Main Methods:

• The algorithm involves two main steps: bone filling, alignment, and ROI computation, followed by cortical bone segmentation.
• Key segmentation tasks include detecting marrow space and pores, calculating cortical thickness, and identifying endosteal boundaries.
• Digital topology and geometry techniques are generalized for automated segmentation.

Main Results:

• The algorithm achieved 95.1% accuracy in volume agreement with manual segmentation on human MD-CT scans.
• An intraclass correlation coefficient of 0.98 was obtained in cadaveric repeat scans, indicating high reproducibility.
• A pilot study using the algorithm revealed significant gender differences in cortical bone thickness and bone mineral density in young adults.

Conclusions:

• The presented algorithm is highly accurate and reproducible for automated cortical bone segmentation in MD-CT imaging of the distal tibia.
• Cortical bone indices derived from this method can quantify gender differences in bone properties.
• Further application to larger populations, including those with compromised bone, is recommended.