Accuracy and applicability of dual-energy computed tomography in quantifying vertebral bone marrow adipose tissue compared with magnetic resonance imaging

Abstract

Abstract Objectives To evaluate the accuracy of dual-energy computed tomography (DECT) in quantifying bone marrow adipose tissue (BMAT) and its applicability in the study of osteoporosis (OP). Methods A total of 83 patients with low back pain (59.77 ± 7.46 years, 30 males) were enrolled. All patients underwent lumbar DECT and magnetic resonance imaging (MRI) scanning within 48 h, and the vertebral fat fraction (FF) was quantitatively measured, recorded as DECT-FF and MRI-FF. A standard quantitative computed tomography (QCT) phantom was positioned under the waist during DECT procedure to realize the quantization of bone mineral density (BMD). The intraclass correlation coefficient (ICC) and Bland–Altman method was used to evaluate the agreement between DECT-FF and MRI-FF. The Pearson test was used to study the correlation between DECT-FF, MRI-FF, and BMD. With BMD as a gold standard, the diagnostic efficacy of DECT-FF and MRI-FF in different OP degrees was compared by receiver operating characteristic (ROC) curve and DeLong test. Results The values of DECT-FF and MRI-FF agreed well (ICC = 0.918). DECT-FF and MRI-FF correlated with BMD, with r values of −0.660 and −0.669, respectively (p < 0.05). In the diagnosis of OP and osteopenia, the areas under curve (AUC) of DECT-FF was, respectively, 0.791 and 0.710, and that of MRI-FF was 0.807 and 0.708, and there was no significant difference between AUCs of two FF values (with Z values of 0.503 and 0.066, all p > 0.05). Conclusion DECT can accurately quantify the BMAT of vertebrae and has the same applicability as MRI in the study of OP.