A comparative analysis of manual glenoid version measurement using two-dimensional and three-dimensional computed tomography imaging techniques.

BACKGROUND

Accurate measurement of glenoid version is essential for optimal implant positioning in shoulder arthroplasty. This study compared the accuracy and reliability of unformatted two-dimensional computed tomography (2D-CT), formatted 2D-CT, and three-dimensional CT (3D-CT) reconstructions in measuring glenoid version.

METHODS

Shoulder CT scans for arthroplasty were analyzed retrospectively. Glenoid version was measured at the estimated glenoid midpoint using unformatted and formatted 2D-CT in the scapular plane. Measurements from 3D-CT reconstructions using the Corrected Friedman Method were used as the control. Inter- and intra-observer reliability was calculated as well as minimally detectable difference.

RESULTS

Sixty-five CT scans were analyzed (mean age, 61.7 years). Mean glenoid version was -3.48° (standard deviation [SD], 8.7°) on unformatted 2D-CT, -3.27° (SD, 8.15°) on formatted 2D-CT, and -4.25° (SD, 7.92°) on 3D-CT. Although no significant difference in mean values was observed (analysis of variance, P=0.245), formatted 2D-CT measurements were within 6° of 3D-CT in 95.4% of cases versus 83.1% for unformatted 2D-CT (P=0.023). Directional error occurred in 27.7% of unformatted scans and 16.9% of formatted scans. Inter-observer reliability was highest for 3D-CT (intraclass correlation coefficient [ICC]=0.83; 95% CI, 0.74-0.89), and intra-observer agreement was strongest for 3D-CT (ICC=0.91; 95% CI, 0.86-0.94), followed by formatted 2D-CT (ICC=0.83; 95% CI, 0.73-0.89) and unformatted 2D-CT (ICC=0.77; 95% CI, 0.65-0.85).

CONCLUSIONS

3D-CT reconstructions are widely considered the most accurate and reproducible method for glenoid version assessment, supported by multiple comparative imaging studies. Formatted 2D-CT provides a reliable alternative when 3D-CT is unavailable, significantly outperforming unformatted 2D-CT in both agreement with the 3D reference and intra- and inter-observer reliability. Level of evidence: IV.