Use of 3D MR reconstructions in the evaluation of glenoid bone loss: a clinical study

Abstract

To assess the ability of 3DMR shoulder reconstructions to accurately quantify glenoid bone loss in the clinical setting using findings at the time of arthroscopy as the gold standard. Retrospective review of patients with MR shoulder studies that included 3DMR reconstructions (3DMR) produced using an axial Dixon 3D-T1W-FLASH sequence at our institution was conducted with the following inclusion criteria: history of anterior shoulder dislocation, arthroscopy (OR) performed within 6months of the MRI, and an estimate of glenoid bone loss made in the OR using the bare-spot method. Two musculoskeletal radiologists produced estimates of bone loss along the glenoid width, measured in mm and %, on 3DMR using the best-fit circle method, which were then compared to the OR measurements. There were a total of 15 patients (13men, two women; mean age, 28, range, 19-51years). There was no significant difference, on average, between the MRI (mean 3.4mm/12.6%; range, 0-30%) and OR (mean, 12.7%; range, 0-30%) measurements of glenoid bone loss (p = 0.767). A 95% confidence interval for the mean absolute error extended from 0.45-2.21%, implying that, when averaged over all patients, the true mean absolute error of the MRI measurements relative to the OR measurements is expected to be less than 2.21%. Inter-reader agreement between the two readers had an IC of 0.92 and CC of 0.90 in terms of percentage of bone loss. 3DMR reconstructions of the shoulder can be used to accurately measure glenoid bone loss.