Publication Details

AFRICAN RESEARCH NEXUS

SHINING A SPOTLIGHT ON AFRICAN RESEARCH

medicine

Comparison of radiograph-based texture analysis and bone mineral density with three-dimensional microarchitecture of trabecular bone

Medical Physics, Volume 38, No. 1, Year 2011

Purpose: Hip fracture is a serious health problem and textural methods are being developed to assess bone quality. The authors aimed to perform textural analysis at femur on high-resolution digital radiographs compared to three-dimensional (3D) microarchitecture comparatively to bone mineral density. Methods: Sixteen cadaveric femurs were imaged with an x-ray device using a C-MOS sensor. One 17 mm square region of interest (ROI) was selected in the femoral head (FH) and one in the great trochanter (GT). Two-dimensional (2D) textural features from the co-occurrence matrices were extracted. Site-matched measurements of bone mineral density were performed. Inside each ROI, a 16 mm diameter core was extracted. Apparent density (Dapp) and bone volume proportion (BV/ TVArch) were measured from a defatted bone core using Archimedes' principle. Microcomputed tomography images of the entire length of the core were obtained (Skyscan 1072) at 19.8 μm of resolution and usual 3D morphometric parameters were computed on the binary volume after calibration from BV/ TVArch. Then, bone surface/bone volume, trabecular thickness, trabecular separation, and trabecular number were obtained by direct methods without model assumption and the structure model index was calculated. Results: In univariate analysis, the correlation coefficients between 2D textural features and 3D morphological parameters reached 0.83 at the FH and 0.79 at the GT. In multivariate canonical correlation analysis, coefficients of the first component reached 0.95 at the FH and 0.88 at the GT. Conclusions: Digital radiographs, widely available and economically viable, are an alternative method for evaluating bone microarchitectural structure. © 2011 American Association of Physicists in Medicine.
Statistics
Citations: 29
Authors: 7
Affiliations: 4
Identifiers
Research Areas
Health System And Policy
Study Approach
Quantitative