Comparative accuracy of lower limb bone geometry determined using MRI, CT, and direct bone 3D models

Area of Science:

• Biomedical Imaging
• Medical Device Technology
• Orthopedic Biomechanics

Background:

• Three-dimensional (3D) bone modeling is crucial for preoperative planning and surgical registration.
• Computer Tomography (CT) provides high bone-soft tissue contrast but involves ionizing radiation.
• Magnetic Resonance Imaging (MRI) is radiation-free but faces challenges with geometric distortion and poor bone contrast, potentially affecting model accuracy.

Purpose of the Study:

• To assess the accuracy of 3D femur and tibia models generated using "Black Bone" 3T MRI compared to high-resolution CT scans.
• To evaluate the geometric validity of MRI-derived bone models against a structured light 3D scanner reference.

Main Methods:

• 12 intact cadaveric lower limbs were scanned using high-resolution CT and "Black Bone" 3T MRI.
• De-fleshed bones were scanned using a structured light (SL) 3D scanner (Polyga HDI COMPACT C210) for reference.
• 3D models were created via image segmentation; CT and MRI models were aligned to SL models using the Iterative Closest Point (ICP) algorithm.
• Differences were quantified using ICP error and Hausdorff distance.

Main Results:

• CT models showed ICP errors of 0.82 ± 0.2 mm (tibia) and 0.85 ± 0.2 mm (femur) compared to SL scans.
• MRI models exhibited ICP errors of 0.97 ± 0.2 mm (tibia) and 0.98 ± 0.18 mm (femur).
• Statistical analysis revealed no significant differences in ICP error or Hausdorff distance between CT, MRI, and SL scanning methods (p > .05).

Conclusions:

• "Black Bone" MRI technique yields accurate 3D geometric measurements of the femur and tibia.
• The accuracy of MRI-derived models is comparable to those obtained with CT.
• The radiation-free nature of MRI presents significant advantages for clinical applications and research in orthopedic imaging.