The aim of the present study was to evaluate the performance of sagittal MR localizer (MR-loc), in terms of diagnostic accuracy and intra- and inter-observer agreement in the detection of vertebral fractures (VFs). Three-hundred MR examinations of the thoracic and/or lumbar spine were randomly collected. A semi-quantitative approach was used and morphometric analysis was performed when a VF was suspected. MR-loc images were evaluated blindly by three radiologists in two different sessions. A full diagnostic sagittal T1-weighted fast spin echo MR sequence was used as standard of reference (RS). Degenerative arthritis was also scored on RS. Only vertebral bodies which were assessable by both MR-loc and RS were considered for the analysis. Area under the receiver operating characteristic curve (AUROC), Cohen kappa statistic, and linear-by-linear association were used for statistical analysis. Kappa values were compared by means of the z distribution. A total of 2186 vertebrae were analysed in 300 MRI exams (147 males, 153 females, 59.4±16.4y.o.). Sixty-seven out of 2136 (3.1%) VFs were identified in 23/300 (7.7%) patients submitted to MRI. In the detection of VFs, sensitivity and specificity of MR-loc were both 100% (accuracy AUROC=1.000). Inter-observer agreement was excellent (k=0.938±0.013), while intra-observer agreement was perfect (k=1.000). The diagnostic performance was independent from degenerative arthritis, vertebral level, type and grade of VFs. MR-loc is a simple but accurate tool in the detection of VFs. It should be introduced for systematic evaluation in the detection of VFs in MR examinations performed in daily clinical practice.
Keywords: Fracture prevention; Osteoporosis; Radiology; Screening.
Copyright © 2014 Elsevier Inc. All rights reserved.