Implementing automated 3D measurements to quantify reference values and side-to-side differences in the ankle...

Area of Science:

• Orthopedics
• Radiology
• Biomedical Engineering

Background:

• Syndesmotic ankle instability detection is challenged by 2D measurement limitations.
• Automated 3D measurement techniques offer potential but lack normative data.
• Normative and side-to-side comparative data for 3D ankle syndesmosis measurements are missing.

Purpose of the Study:

• Establish 3D anatomical reference values for the ankle syndesmosis using automated measurements.
• Determine the degree of ankle syndesmosis symmetry across all 3D measurements.

Main Methods:

• Retrospective inclusion of patients without syndesmotic pathology undergoing CT scans (NWBCT and WBCT).
• Training a neural network for automated 3D ankle model segmentation.
• Iterative closest point registration for left-to-right ankle superimposition and 3D measurement computation.

Main Results:

• Excellent intra-observer agreement for manual measurements (ICC 0.85-0.99).
• Good accuracy between automated and manual measurements (angles <2.7°, distances <0.5 mm).
• High accuracy for automated segmentation (Dice coefficient 0.99) and established 3D reference values.

Conclusions:

• Automated 3D measurements provide reliable reference values for ankle syndesmosis.
• Ankle syndesmosis demonstrates significant side-to-side symmetry.
• The novel algorithm can overcome 2D measurement limitations for syndesmotic lesion detection.