Cross-calibration of a fan-beam X-ray densitometer with a pencil-beam system

This study describes the cross-calibration of two Hologic bone densitometers, one using a fan-beam X-ray source (QDR4500/A) and the other a pencil-beam source (QDR1000/W). The QDR4500/A allows spine and hip measurements to be made at three principal speeds. Results from two spine phantoms and 154 patients showed no significant difference in the absolute values or precision between speeds. The middle speed (taking 1 min to scan an adult lumbar spine) was used for the comparison between the instruments. The two densitometers were compared using two spine phantoms and 182 patients. In vivo measurements were made of all lumbar spine, hip, forearm and whole body sites. Regression lines, constrained to pass through the origin, were calculated. Slopes for total bone mineral density (BMD) for each scan type ranged between 0.994 and 1.029, the best value being found for forearm (1.000). Scatter graphs of the individual points were generated and showed results slightly worse than would be expected from repeat measurements on a single machine (79-88% fell within the expected 2 SD range). A trend for the QDR4500/A to overestimate BMD at low values and underestimate it at high values was seen in the femoral neck. The trend was more significant in the lumbar spine. There was an overestimate of total hip BMD throughout the range. Slopes of the regression lines for area and bone mineral content (BMC) were used to improve cross-calibration between the systems on a site-to-site basis, after which the results improved to a level consistent with repeat measurements on a single machine (81-94% within 2 SD). At present only global, rather than site-specific, correction factors can be employed for the spine and hip and no overall improvement in cross-calibration was possible. We conclude that although global correction factors allow adequate cross-calibration to be achieved, improvements could be made by the use of scan-site-specific factors.