Further improvements on the factors affecting bone mineral density measured by quantitative micro-computed tomography

The effects of imaging parameters and special configuration of objects within the reconstruction space on the micro computed tomography (mu CT) based mineral density have been explored, and a series of density correction curves have been presented. A manufacturer-provided calibration phantom (0, 100, 200, 400, 800 mg HA/cm(3)) was imaged at all possible imaging conditions (n = 216) based on energy, resolution, vial diameter, beam hardening correction factor and averaging. For each imaging condition, a linear regression model was fitted to the observed versus expected densities, and the intercepts (beta(0)) and slopes (beta(1)) of the regression lines and each density level were modeled using multiple regression modeling. Additionally, a custom made phantom (0, 50;150, 500, 800, 1000 and 1500 mg HA/cm(3)) was scanned in order to study the effects of location and orientation of an object within the reconstruction space and presence of surrounding objects on mu CT based mineral density. The energy, vial diameter and beam hardening correction factor were significant predictors of cumineral density (P values<0.001), while averaging and resolution did not have a significant effect on the observed density values (P values>0.1) except for 0.0 density (P values<0.04). Varying the location of an object within the reconstruction space from the center to the periphery resulted in a drop in observed mineral density up to 10% (P values<0.005). The presence of surrounding densities resulted in decreased observed mineral density up to 17% at the center and up to 14% at the periphery of the reconstruction space (P values<0.001 for all densities). Changing the orientation of the sample also had a significant effect on the observed mineral density, resulting in up to 16% lower observed mineral density for vertical vs. horizontal orientation at the center of the reconstruction space (P value<0.001). We conclude that energy, resolution and post processing correction factor are significant predictors of the observed mineral density in mu CT. (C) 2011 Elsevier Inc. All rights reserved.