Site-Specific Associations Between CBCT-Derived Gray Value and Fractal Dimension and DEXA T-Scores in Postmenopausal Women: A Cross-Sectional Analytical Study

Osteoporosis is a silent systemic skeletal disease characterized by microarchitectural deterioration and increased fracture risk; thus, early case-finding is essential. Although dual-energy X-ray absorptiometry (DEXA) is the reference standard for bone mineral density (BMD) assessment, limited accessibility and cost restrict its use for large-scale screening. Given that cone-beam computed tomography (CBCT) is routinely obtained in dental practice, image-derived parameters-gray value (GV) and fractal dimension (FD)-may provide insight into their association with systemic bone mineral density measured by DEXA in postmenopausal women, without additional radiation exposure.

This cross-sectional study included 63 postmenopausal women aged 50-70 years undergoing dental CBCT (NewTom 3G; standardized acquisition). Six trabecular regions of interest (ROIs) were evaluated: mandibular retromolar, mandibular premolar, mandibular anterior (central), maxillary premolar, maxillary anterior (central), and maxillary tuberosity. GV values were extracted using OnDemand3D-Dental from standardized 25 × 25-pixel ROIs positioned to exclude cortical bone and dental structures. FD was calculated in ImageJ using a box-counting method following Gaussian blur subtraction, binarization, and skeletonization. Lumbar spine and hip BMD were measured by DEXA and categorized according to WHO criteria. Statistical analyses included t-tests/ANOVA, Pearson correlation (α = 0.05). Intra-observer reliability was assessed with intraclass correlation coefficients (ICC).

DEXA classified 49.2% as normal, 34.9% as osteopenic, and 15.9% as osteoporotic. Lumbar assessment identified more non-normal cases than the hip evaluation. Reliability was good (ICC > 0.70). Although mean GV and FD did not differ significantly between groups, site-specific GV-FD correlations were observed. Notably, FD at the maxillary premolar correlated with lumbar T-scores in the non-normal group (p = 0.030).

While GV and FD did not discriminate BMD categories overall, region-dependent associations between CBCT-derived parameters and systemic BMD were observed, suggesting potential clinical relevance in dental imaging. These exploratory findings should not be interpreted as diagnostic or screening indicators and require confirmation in larger studies.