Mineral density and bone architecture properties are the main measures of bone quality. Dual-energy X-ray Absorptiometry (DXA) is the traditional clinical measurement technique for bone mineral density, but it is insensitive to architectural information. Image analysis of the architectural properties of bones can be used to predict bone quality. This study is aimed at investigating the statistical parameters extracted from wo dimensional projection images of the DXA scans and exploring its link with architectural properties, and its correlation with a bones mechanical properties. In this research, features extracted from the Gray Level Co-occurrence Matrix (GLCM) for a 2D image are compared with features extracted from semivariogram analysis in order to estimate bone micro-architectural and mechanical properties. Data analysis was conducted on 13 trabecular bones of different strengths (with an in-plane spatial resolution of about 50Î¼m). Ground truth data for bone volume fraction (BV/TV), bone strength and elasticity was available for the dataset, based on complex 3D analysis and mechanical tests. Correlation between the statistical parameters and biomechanical test results was studied using regression analysis. The results showed that the cluster-shade parameter extracted from the GLCM was strongly correlated with the microstructure of the trabecular bone and also somewhat related to the mechanical properties. Additionally, a parameter called 'sill' obtained by the semivariogram method was found to be highly associated with the mechanical properties of the bone and slightly related to its microarchitectural properties.
Date of publication
Huang, Ning, "The Measurement of Bone Quality in Medical Images Using Statistical Textural Features" (2011). Electrical Engineering Theses. Paper 3.