Influence of Radiographic Viewing Perspective on Glenoid Inclination Measurement

Abstract

Introduction The purposes of this study were to determine (1) whether glenoid inclination (GI) could be accurately measured on plain radiographs as compared to a gold-standard 3-dimensional (3D) measure and (2) whether GI could be reliably measured on plain radiographs. Materials and Methods Digitally reconstructed radiographs (DRRs) were made from 3D computed tomography reconstructions of 68 normal cadaver scapulae. DRRs were made in a variety of viewing angles. Inclination was measured on these DRRs. These measurements were also made using a gold-standard 3D method. Measurements were made by 2 orthopedic surgeons and 1 surgeon twice, to calculate interrater and intrarater intraclass correlation coefficients (ICCs). Results The gold-standard 3D β was 83 ± 5° (72°–98°). On neutral plain radiographs, the mean ± standard deviation 2D β angle was 80 ± 6° (range, 66°–99°). With regard to accuracy, the 2D β angle was significantly different from the 3D β angle, with the 2D β underestimating the 3D β by 5° (95% confidence intervals −1 to 12). With regard to reliability, interrater ICCs for 2D β with a neutral viewing angle was 0.79. Two-dimensional β varied widely with viewing angle from 0.24 to 0.88. Interrater ICCs for the 3D method was 0.83 (0.60–0.92). Intrarater ICCs for all 3 techniques were high (>0.91). Conclusions Two-dimensional radiographic GI measurement is not accurate, as it underestimates the 3D value by an average of 5° when compared to the gold-standard 3D measurement. GI 2D measurement reliability varies with viewing angle on plain radiographs and thus to accurately and reliably measure inclination 3D imaging is necessary.