Fractal analysis of proximal femur radiographs: Correlation with biomechanical properties and bone mineral density

Conventional radiography and fractal analysis were used to quantify trabecu lar texture patterns in human femur specimens and these measures were used in conjunction with bone mineral density (BMD) to predict bone strength. Ra diographs were obtained from 51 human femur specimens (25 male, 26 female). The radiographs were analyzed using three different fractal geometry based techniques, namely semi-variance, surface area and Fourier analysis. Maxim um compressive strength (MCS) and shear stress (MSS) were determined with a material testing machine; BMD was measured using quantitative computed tom ography (QCT). MCS and MSS both correlated significantly with BMD (MCS: R = 0.49-0.54; MSS: R = 0.69-0.72). Fractal dimension also correlated signific antly with both biomechanical properties (MCS: R = 0.49-0.56; MSS: R = 0.47 -0.54). Using multivariate regression analysis, the fractal dimension in ad dition to BMD improved correlations versus biomechanical properties. Both B MD and fractal dimension showed statistically significant correlation with bone strength. The fractal dimension provided additional information beyond BMD in correlating with biomechanical properties.