Methods for Computed Tomography-Based Glenoid Bone Loss Measurement Show High Interobserver Consistency but Variable Accuracy

To evaluate the reliability of computed tomography (CT)-based glenoid bone loss calculation methods in shoulder instability by assessing interobserver agreement using 3-dimensional virtual bone models and to identify the technique that provides the most optimal balance for clinical application.

Between January 2023 and June 2023, normal shoulder CT scans without evidence of instability or prior surgery were retrieved from institutional archives. Virtual anteroinferior glenoid bone defects of predefined sizes (10%, 15%, 20%, 25%, 30%, 35%, and 40%) were created on a standardized en face glenoid view using CT console-based segmentation tools, establishing a controlled reference standard. Four senior musculoskeletal radiologists independently measured glenoid bone loss on both 2-dimensional and 3-dimensional reconstructions using 4 established techniques (Sugaya, PICO, Hamamoto, and Parada). Interobserver reliability was assessed using the intraclass correlation coefficient (ICC), and agreement with the predefined defect size was evaluated using Bland-Altman analysis.

The PICO method showed the highest interobserver reliability (ICC = 0.995) but showed systematic overestimation with the largest mean bias (+6.93 mm) and wide limits of agreement, reflecting proportional bias. The Sugaya method showed both high reproducibility (ICC = 0.854) and minimal bias (-1.06 mm), making it the most balanced technique. The Hamamoto method showed good reliability (ICC = 0.952) but a tendency to overestimate (+4.04 mm). The Parada method showed poor reproducibility (ICC = 0.223) despite moderate accuracy (-2.85 mm).

The Sugaya method showed the most favorable combination of accuracy and interobserver reliability for CT-based glenoid bone loss measurement in a controlled virtual model. The PICO method, despite excellent interobserver consistency, showed systematic overestimation. The Hamamoto method showed acceptable reliability with moderate overestimation, whereas the Parada method showed poor reproducibility.

Reliable glenoid bone loss estimation is vital for surgical planning in shoulder instability. According to our study, the Sugaya method offers the most accurate and reproducible measurements, helping surgeons make better-informed decisions and avoid inappropriate procedures.