Correction of the effects of source, source strength, and soft-tissue thickness on spine dual-photon absorptiometry measurements

Summary

Dual photon absorptiometry measurements of the spine are subject to drift associated with source, source strength, and truncal thickness. This study was conducted to determine the extent to which this drift in bone mineral density (BMD) measurements can be improved by analysis of scans with a new software version, 08C, and by applying external standard or phantom corrections to scans analyzed with the older version, 08B. A phantom, consisting of human lumbar vertebrae embedded in acrylic, and five clear acrylic plates to simulate a soft-tissue thickness range of 15.2–27.9 cm, was measured on a Lunar Radiation Corp DP3 scanner over the life of a 153-gadolinium (Gd) source and scans analyzed with software versions 08B and 08C. Phantom BMD was lower with 08C at both high [0.012±0.002 (SEM) g/cm²,P<0.001] and low (0.027±0.003 g/cm²,P<0.001) count rates than with 08B. Phantom BMD of scans analyzed with 08B increased with increasing source age and the source strength-related increment increased significantly as acrylic thickness increased (P=0.014). When the same scans were analyzed with 08C, the thickness-related effect was corrected whereas a small (0.011 g/cm²/year) source-strength effect persisted. The effects of source strength and truncal thickness on BMD were also evaluated in 40 humans scanned at two detector collimations to vary count rate. With 08B, mean BMD was 1% greater when measured with 8 than with 13 mm collimation (mean difference 0.011±0.003 g/cm²,P=0.001), whereas the version 08C, mean BMD was the same at the two collimations. Similarly, when phantom corrections were applied to the scans analyzed with 08B, the source strength effect was no longer significant. A truncal thickness effect, apparent in the 40 human scans analyzed with 08B, was not present with 08C. Finally, the phantom was scanned with three different Gd sources. With both 08B and 08C, BMD values were similar with two and significantly lower (by 0.012±0.002 g/cm²,P=0.011) with the third Gd source. Thus, with the new analysis software 08C, multiple thickness calibration is no longer needed, however, calibration with an external standard is still necessary.