Variability, agreement and reliability of MRI knee landmarks.

Area of Science:

• Medical Imaging
• Orthopedic Biomechanics
• Anatomical Modeling

Background:

• Surface mesh reconstructions of knee bones from MRI scans are vital for defining coordinate systems and regions of interest.
• Understanding the variability, agreement, and reliability of osseous landmarks is essential for accurate soft tissue analysis.

Purpose of the Study:

• To quantify the variability, agreement, and reliability of osseous landmarks on distal femur and proximal tibia surface meshes derived from MRI.
• To assess the downstream effects of landmark placement accuracy on knee soft tissue studies.

Main Methods:

• Defined 15 landmarks on the distal femur and 12 on the proximal tibia from 20 knee MRI scans.
• Created surface meshes and performed landmark identification by single and multiple observers with repeated measures.
• Utilized Procrustes analysis for shape alignment, principal component analysis for variability, and two-way ANOVA for agreement and reliability.

Main Results:

• Inter-subject landmark variation ranged from 0.6 to 5.26 mm.
• Inter-observer agreement was at most 5.1 mm, and intra-observer agreement was at most 5.69 mm.
• Observer reliability varied significantly (0.07-0.98 between-observer, 0.51-0.98 within-observer), with sphere/circle-derived landmarks often performing better.

Conclusions:

• Osseous landmark placement on knee MRI-derived meshes exhibits quantifiable variability and observer-dependent agreement.
• Reliability metrics are crucial for interpreting downstream analyses in knee biomechanics and soft tissue studies.
• Specific landmark definitions (e.g., fitted spheres/circles) may offer improved consistency.