Evaluation of Bone Mineral Density: Correlating MRI Cervical Vertebral Bone Quality, CT Hounsfield Units, and DEXA T-Scores.

BACKGROUND/OBJECTIVES: Dual-energy X-ray absorptiometry (DEXA), the gold standard for assessing bone mineral density (BMD), may yield inaccurate results in certain populations. This has prompted interest in alternative imaging methods, including the MRI-based cervical and lumbar vertebral bone quality (CVBQ and LVBQ) scores. The lumbar VBQ score is a validated MRI-based metric with excellent inter- and intra-rater reliability and established clinical utility in preoperative spine assessment, whereas the newer cervical VBQ (CVBQ) score has shown mixed results in early studies. This study investigates associations between a novel CVBQ score derived from MRI and established BMD metrics (Hounsfield unit (HU) measurements and DEXA values) to evaluate the comparative utility of these methods. METHODS: A retrospective review was performed on patients who underwent cervical CT, DEXA, and non-contrast MRI of the cervical and lumbar spine between 2016 and 2022. BMD was assessed using DEXA T-scores, cervical HU measurements, and CVBQ and LVBQ scores derived from T1-weighted MRI sequences. Statistical comparisons between patients with and without osteoporosis were conducted using t-tests and Pearson correlation coefficients. RESULTS: A total of 133 patients were included for CVBQ scoring and 127 for LVBQ. The CVBQ score showed poor correlation with both DEXA (r = -0.09, p < 0.001) and HU measurements (r = -0.34, p < 0.001), whereas a moderate correlation was found between LVBQ and DEXA scores (r = -0.34, p < 0.001). CONCLUSIONS: The LVBQ score demonstrates moderate correlation with DEXA and may serve as a reliable tool for preoperative BMD assessment. However, the CVBQ score showed weak correlation with both DEXA and HU, limiting its clinical utility in its current form. Further refinement of the CVBQ methodology is needed to enhance its accuracy and relevance for surgical risk assessment and postoperative outcome prediction.