Cortical shell unwrapping for vertebral body abnormality detection on computed tomography

Area of Science:

• Orthopedics
• Radiology
• Medical Imaging

Background:

• Vertebral body assessment for acute injury and chronic deformity is challenging via visual inspection.
• Accurate quantitative analysis of vertebral body pathologies remains a significant clinical need.

Purpose of the Study:

• To develop and evaluate a cortical shell unwrapping method for quantitative assessment of vertebral body injuries.
• To improve the accuracy and efficiency of detecting fractures and degenerative osteophytes in the spine.

Main Methods:

• Spine segmentation and partitioning into individual vertebrae.
• Extraction of the vertebral body cortical shell using deformable dual-surface models.
• Unwrapping the 3D cortical shell onto a 2D map for pattern recognition using support vector machines.

Main Results:

• Fracture detection Computer-Aided Detection (CAD) achieved 93.6% sensitivity with 3.2 false positives per patient on CT datasets.
• Degenerative osteophyte CAD achieved 82% sensitivity with 4.7 false positives per patient on PET/CT datasets.

Conclusions:

• The proposed cortical shell unwrapping method effectively transforms 3D vertebral body analysis into a 2D pattern recognition problem.
• This approach shows significant potential for improving the diagnosis of vertebral fractures and degenerative changes.