Bone texture characterization by oriented fractal analysis

The Hurst coefficient H can be interpreted as a measure of the texture contained within an image. In this work, the analysis of texture is applied to bone sections in an attempt to quantify calcification. Because the bone structure is generally nonisotropic, a radially oriented implementation of a maximum likelihood estimator for the Hurst coefficient H is proposed to determine a directionally dependent measure of texture. This method is applied to two sets of X-ray images; excised, bisected femoral heads dissolved for various periods in an acid bath, and an immobilized human heel (calcaneus) after an injury. The results show that the value of H can be mapped to the visual evolution of bone structure, and that H decreases with the dissolution of calcium. To evaluate the robustness of the classification, different conditions of image acquisition were tested.<
>