Related Experiment Video
Updated: May 9, 2025
10:23
Author Spotlight: Three-Dimensional Cephalometric Landmark Annotation Demonstration on Human Cone Beam Computed Tomography Scans
Published on: September 8, 2023
2.5K
Semiautomated Three-Dimensional Landmark Placement on Knee Models Is a Reliable Method to Describe Bone Shape and
Nancy Park1, Johannes Sieberer1, Armita Manafzadeh1
1Yale University, New Haven, Connecticut, U.S.A.
Arthroscopy, Sports Medicine, and Rehabilitation
|April 29, 2025
Summary
Artificial intelligence and human input demonstrated excellent reliability in identifying knee anatomical landmarks. This supports using AI-assisted 3D methods for precise knee measurements in orthopaedic research.
Area of Science:
- Orthopaedic research
- Medical imaging
- Artificial intelligence
Background:
- Artificial intelligence (AI) is increasingly explored in medical imaging and orthopaedic research.
- Accurate anatomical landmark identification is crucial for precise knee measurements.
- Current 2D methods may lack the precision of 3D approaches.
Purpose of the Study:
- To evaluate the inter- and intrarater reliability of 21 anatomical knee landmarks.
- To assess landmarks initially placed by an AI algorithm and manually verified.
- To compare AI-assisted methods with traditional manual measurements.
Main Methods:
- Thirty knee CT scans from the Multicenter Osteoarthritis Study (MOST) were analyzed.
- An AI algorithm automatically identified 19 landmarks; 2 were added manually.
- Two reviewers independently verified landmarks, with one repeating the process after 2 weeks.
Main Results:
- Excellent inter-rater reliability was observed for all 21 landmarks.
- Intraclass correlation coefficients (ICCs) ranged from 0.87 to 1.00 for inter-rater reliability.
- Intrarater reliability showed ICCs between 0.90 and 1.00 for most landmarks.
Conclusions:
- High agreement was achieved for all 21 evaluated anatomical landmarks.
- AI-assisted landmark identification shows strong reliability.
- Semiautomated 3D methods offer enhanced precision for knee anatomical measurements.
