Validation and accuracy evaluation of automatic segmentation for knee joint pre-planning.

Area of Science:

• Orthopedic surgery
• Medical imaging analysis
• Computational anatomy

Background:

• Accurate 3D models from medical imaging are crucial for clinical applications like preoperative planning.
• The reliability of patient-specific 3D anatomical reconstructions must be proven before widespread clinical adoption.

Purpose of the Study:

• To assess the dimensional accuracy of automatically segmented 3D knee joint models compared to manually segmented models.
• To validate the use of automatic segmentation for generating patient-specific 3D anatomical models.

Main Methods:

• Three-dimensional models of the femur and tibia were created from CT scans using both manual and automatic segmentation.
• Key bony landmarks were identified to measure clinically relevant distances.
• Pearson's correlation and Bland Altman plots analyzed the agreement between measurements from both segmentation methods.

Main Results:

• Differences in measured distances between automatic and manual segmentation were generally below 1 mm.
• A strong correlation was observed between measurements from both segmentation techniques.
• The tibial knee center tubercle distance (TKCTD) showed slightly larger differences in two specimens.

Conclusions:

• Automatic segmentation provides reliable 3D bone models comparable in accuracy to manual segmentation.
• This technology can accelerate the creation of 3D anatomical models for clinical use.
• The findings support the increased adoption of automatic segmentation in preoperative planning and surgical settings.