The Adriamycin cost outer surface was then eroded by 3 pixels to return the ROI boundary approximately to the periosteal edge. Following alignment in the common coordinate system, the grayscale images were spatially masked using the radius periosteal VOI. In this manner, the ulna and all extra-osseal soft tissue did not contribute to the projected image, approximating the soft tissue compensation inherent to DXA. The masked 3D image was then projected along the dorsal–palmar direction (y′-axis) according to the discrete line integral: $$ \textaBMD_\textsim \left( x\prime, z\prime \right) = \sum\limits_y\prime = 1^y\prime = N \left[ \textHA \right]\left( x\prime, y\prime, z\prime \right)\Delta

y $$ (1)where aBMDsim is the simulated areal bone mineral density of the distal radius projected onto the x′z′-plane (corresponding to medial–lateral and superior–inferior axes), [HA](x′,y′,z′) is the aligned 3D HR-pQCT-calibrated mineral density image matrix, N is the number of voxels in the y′ direction, and Δy is the voxel size in y′. The mean aBMDsim was then calculated as the arithmetic average of all non-zero pixels from

this projected image. Reproducibility Reproducibility of the aBMDsim measurement was see more determined in 8 radii of volunteers spanning a large age range (age = 25 to 65 years). Three repeat measurements were performed for each subject with complete repositioning between each scan. For three of the patients, a single dataset was excluded due to excessive motion artifacts visually apparent in the reconstructed images. Therefore, a total of five patients with three scans and three patients with two scans were used to calculate the root mean squared coefficient of variation Ribonuclease T1 (RMS-CV%) for aBMDsim. DXA Areal bone densitometry data were acquired for the radius, proximal femur, and lumbar spine using one of two commercial DXA scanners; 42 osteopenic women from the first cohort were scanned with the QDR 4500 (Hologic Inc., Bedford, MA, USA) and the remaining 75 subjects were scanned using the Lunar Prodigy (GE Healthcare, Chalfont St. Giles, UK).

Standard ROIs used for clinical assessment of osteoporosis status were identified to determine aBMD. The UD region of interest was automatically determined by the scanner software (Fig. 1b, c). For the Hologic device, this region started at the most proximal end of the endplate of the radius and extended 15 mm proximally. For the Lunar device, the region started where the radius and ulna superimpose and extended proximally for 20 mm. Mean BMD values from the UD ROI will subsequently be referred to as aBMDdxa. Areal BMD measures were also determined for the lumbar spine (L1–L4) and total proximal femur using the standard densitometry protocols and analysis software provided by the manufacturer. Statistics Mean and standard deviations were calculated for all indices.