Three-Dimensional Magnetic Resonance Imaging FRACTURE (Fast Field Echo Resembling A CT Using Restricted Echo-Spacing) Sequence Is Equivalent to Three-Dimensional Computed Tomography in Quantifying Bone Loss and Measuring Shoulder Morphology in Patients With Shoulder Dislocation

Abstract

Purpose: To evaluate the equivalence of 3-dimensional (3D) magnetic resonance imaging (MRI) (FRACTURE [Fast field echo Resembling A CT Using Restricted Echo-spacing]) and 3D computed tomography (CT) in quantifying bone loss in patients with shoulder dislocation and measuring morphologic parameters of the shoulder.

Methods: From July 2022 to June 2023, patients with anterior shoulder dislocation who were aged 18 years or older and underwent both MRI and CT within 1 week were included in the study. The MRI protocol included an additional FRACTURE sequence. Three-dimensional reconstructions of MRI (FRACTURE) and CT were completed by 2 independent observers using Mimics software (version 21.0) through simple threshold-based segmentation. For bone defect cases, 2 independent observers evaluated glenoid defect, percentage of glenoid defect, glenoid track, Hill-Sachs interval, and on-track/off-track. For all cases, glenoid width, glenoid height, humeral head-fitting sphere radius, critical shoulder angle, glenoid version, vault depth, and post-processing time were assessed. The paired t test was used to assess the differences between 3D CT and 3D MRI (FRACTURE). Bland-Altman plots were constructed to evaluate the consistency between 3D CT and 3D MRI (FRACTURE). Interobserver and intraobserver agreement was evaluated with the interclass correlation coefficient. The paired χ² test and Cohen κ statistic were used for binary variables (on-track/off-track).

Results: A total of 56 patients (16 with bipolar bone defect, 5 with only Hill-Sachs lesion, and 35 without bone defect) were ultimately enrolled in the study. The measurements of 21 bone defect cases showed no statistically significant differences between 3D CT and 3D MRI: glenoid defect, 4.05 ± 1.44 mm with 3D CT versus 4.16 ± 1.39 mm with 3D MRI (P = .208); percentage of glenoid defect, 16.21% ± 5.95% versus 16.61% ± 5.66% (P = .199); glenoid track, 18.02 ± 2.97 mm versus 18.08 ± 2.98 mm (P = .659); and Hill-Sachs interval, 14.29 ± 1.93 mm versus 14.35 ± 2.07 mm (P = .668). No significant difference was found between 3D CT and 3D MRI in the diagnosis of on-track/off-track (P > .999), and diagnostic agreement was perfect (κ = 1.00, P < .001). There were no statistically significant differences between the 2 examination methods in the measurements of all 56 cases, except that the post-processing time of 3D MRI was significantly longer than that of 3D CT: glenoid height, 34.56 ± 1.98 mm with 3D CT versus 34.67 ± 2.01 mm with 3D MRI (P = .139); glenoid width, 25.32 ± 1.48 mm versus 25.45 ± 1.47 mm (P = .113); humeral head-fitting sphere radius, 22.91 ± 1.70 mm versus 23.00 ± 1.76 mm (P = .211); critical shoulder angle, 33.49° ± 2.55° versus 33.57° ± 2.51° (P = .328); glenoid version, -3.25° ± 2.57° versus -3.18° ± 2.57° (P = .322); vault depth, 37.43 ± 1.68 mm versus 37.58 ± 1.75 mm (P = .164); and post-processing time, 89.66 ± 10.20 seconds versus 360.93 ± 26.76 seconds (P < .001). For all assessments, the Bland-Altman plots showed excellent consistency between the 2 examination methods, and the interclass correlation coefficients revealed excellent interobserver and intraobserver agreement.

Conclusions: Three-dimensional MRI (FRACTURE) is equivalent to 3D CT in quantifying bone loss in patients with shoulder dislocation and measuring shoulder morphologic parameters.

Level of evidence: Level II, development of diagnostic criteria (consecutive patients with consistently applied reference standard and blinding).