Association of Radiograph-Derived Femoral Neck Fractal Dimension With Bone Mineral Density in Postmenopausal Women

Abstract

Bone mineral density (BMD) is central to osteoporosis diagnosis but incompletely reflects bone microarchitecture, a key determinant of fracture risk. Fractal analysis of radiographs has been proposed as a low-cost way to characterize trabecular structure, but its relationship with densitometric measures remains uncertain. This study investigated the association between femoral neck fractal dimension (FD) derived from pelvic radiographs and DXA-derived BMD and T-scores in postmenopausal women. This retrospective, cross-sectional study included postmenopausal women who underwent anteroposterior pelvis radiography and DXA within the same year. FD was computed from a standardized femoral neck region of interest using the box-counting method. DXA outcomes were BMD and T-scores at the lumbar spine, femoral neck, and total hip. Associations were evaluated using Spearman correlation and univariable linear regression. A total of 152 women were included; mean femoral neck FD was 1.297 ± 0.083. FD showed no meaningful correlation with DXA-derived measurements (Spearman ρ range, −0.004 to 0.122; all p > 0.05; largest at total hip T-score, ρ=0.122, p=0.134). In univariable regression, FD was not associated with any densitometric endpoint (all p > 0.05), with negligible model fit (R² range, 0.000–0.010). Radiograph-derived femoral neck FD was not associated with DXA BMD or T-scores and provided essentially no explanatory or predictive value for densitometric outcomes. These findings argue against its use as a surrogate for DXA and support the need for standardized methods with prospective outcome validation.