Automated Posterior Tibial Slope Measurement Using Lateral Knee Radiographs: A Novel Landmark-Based Approach Using...

Area of Science:

• Orthopaedic Surgery
• Radiology
• Medical Imaging Analysis

Background:

• A standardized protocol for measuring posterior tibial slope (PTS) is lacking.
• This absence hinders surgical decision-making and risk stratification using established cutoff values.

Purpose of the Study:

• To validate an online computer vision model for PTS measurement on lateral knee radiographs.
• To assess the model's accuracy compared to manual measurements by orthopaedic specialists.

Main Methods:

• Utilized 10,007 lateral knee radiographs for training, validation, and testing.
• Developed two landmark-based methods (short and long) for determining the tibial shaft axis.
• Evaluated model performance using inter- and intraobserver intraclass correlation coefficients (ICCs) and time efficiency.

Main Results:

• Achieved high interobserver ICCs (0.91-0.92) and near-perfect intraobserver ICC (1.00) for the computer vision model.
• Demonstrated excellent accuracy across normal, osteoarthritic, and implant-embedded knees (ICCs 0.84-0.97).
• The model significantly reduced measurement time from 26.1 seconds to 2.5 seconds (P < .001).

Conclusions:

• A novel, time-efficient deep learning model for PTS measurement shows excellent accuracy and consistency.
• This model has the potential to standardize PTS measurements for clinical translation.
• External validation is recommended to confirm its utility in diverse clinical settings.