Can EOS Imaging Substitute for Conventional Radiography in Measurement of Acetabular Morphology in the Young Dysplastic Hip?

EOS imaging offers a low-radiation alternative to conventional radiography (CR) and has little to no magnification effects. However, it is unclear how radiographic measures may be affected using EOS. The present study aims to determine the reproducibility of measures of acetabular morphology on EOS images as compared with CR, and to directly compare the 2 imaging modalities.Methods:A total of 21 consecutive patients (66.7% female; 14.4±4.7 y) indicated for an open hip preservation procedure with both an anterior-posterior pelvis radiograph and EOS image performed preoperatively were included. Three orthopaedic surgeons measured Tönnis angle, lateral center edge angle (LCEA), acetabular depth-width ratio (ADR), and extrusion index (EI). Measurements were performed twice, 2 weeks apart. Reliability between observers and time points was measured using intraclass correlation coefficients, and agreement between time points and modalities was measured using Bland-Altman analysis.Results:On EOS images, inter-rater reliability was 0.86 for Tönnis angle, 0.86 for LCEA, 0.74 for ADR, and 0.93 for EI. On CR, inter-rater reliability was 0.86 for Tönnis anlge, 0.90 for LCEA, 0.82 for ADR, and 0.84 for EI. In the agreement analysis, biases between imaging modalities were observed. On average, raters measured Tönnis angle and EI higher on EOS images than CR (2.22 degrees, 1.09%, respectively); and LCEA and ADR lower on EOS images than CR (1.54 degrees, 1.14% respectively). Limits of agreement (LOA) between modalities were similar to that of LOA observed in intra-rater analysis.Conclusions:Measures of acetabular morphology performed on EOS images have similarly high intra-rater and inter-rater reliability compared with CR. Measures performed on EOS also have similar intra-rater agreement as compared with CR. Intermodality agreement had similar LOA as intra-rater agreement on either individual imaging modality. Small biases between imaging modalities were detected.Levels of Evidence:Level I - diagnostic study: investigating a diagnostic test.