Comparison of two Hologic DXA systems (QDR 1000 and QDR 4500/A).

Publisher Website
Google Scholar
Abstract
Bone mineral content is reliably measured by dual energy X-ray absorptiometry (DXA), if manufacturers' recommendations and quality control (QC) procedures are followed. Several phantoms (Hologic anthropomorphic spine phantom, the Groupe de Recherche et d'Informations sur les Osteoporoses (GRIO) test objects and the European semi-anthropomorphic phantoms) were used to evaluate reproducibility, linearity, accuracy and spatial resolution of two DXA devices in vitro. These parameters were also evaluated in vivo from measurements performed on 120 volunteer patients. It was found that when one device (a single beam monodetector QDR 1000) is replaced by another (a fan beam multidetector QDR 4500/A), the novel combination of procedures described here, ensures that the accuracy of DXA study results is maintained when both devices are used in succession for the same patient. To study the possible responses in clinical situations, the influence of bone environment (soft and adipose tissues) was also evaluated. In both systems, similar performances (in vitro coefficients of variation of 0.5%) were established. At extreme bone density values, slight differences in linearity were found, as well as differences in accuracy and spatial resolution. Lumbar spine and femoral neck measurements were performed with both systems in 120 volunteers, both measurements being made on the same day. The corresponding bone mineral density (BMD) values were highly correlated (r2 = 0.985 for lumbar spine and 0.948 for the femoral neck), and the mean BMD differences were 0.68% and 0.37% for each anatomical site, respectively. Although small, these differences add to the precision error of the method, which is near 1%. A calibration curve has to be obtained in order that both devices can be equally used in regular clinical study. We concluded that when a DXA system is replaced by a new one, appropriate QC procedures must be strictly observed.

This publication has 15 references indexed in Scilit: