A method for the estimation of femoral bone mineral density from variables of ultrasound transmission through the human femur

Quantitative ultrasound (QUS) measurements at peripheral sites can be used to estimate osteoporotic fracture risk. However, measurements at these sites are less suitable to predict bone mineral density (BMD) or fracture risk at the central skeleton. We investigated whether direct QUS measurements at the femur would allow to estimate dual X-ray absorptiometry (DXA) BMD of the total proximal femur with errors comparable to established DXA accuracy errors. Two independent sets of femora were measured in Kiel (6 f, 4 m, age: 55-90) and Paris (19 f, 20 in age: 45-95) using different benchtop systems in the two laboratories. The femora were scanned in transverse transmission mode using focused US transducers of 500 kHz center frequency. The QUS values were averaged over a region similar to the total hip region of dual X-ray absorptiometry (DXA) measurements. BMD was measured using DXA. SOS and BMD correlated significantly (p < 0.0001) in both data sets (R-2=0.81-0.93). Correlations between BUA and BMD were also significant at p < 0.001, but correlation coefficients were lower (R-2 = 0.61-0.75). Residual errors for the estimation of BMD were 8%-10% for SOS as predictor, and 14%-16% for BUA as predictor. The residual error of 8 to 10% for the estimation of BMD from SOS is comparable to variabilities among different DXA femur subregions and accuracy errors of femoral DXA measurements caused by the impact of soft tissue. It is substantially smaller than the errors of 13% for the estimation of total femur BMD from spine BMD, 14% for the estimation of total femur BMD from calcaneus SOS or 16% for the estimation of ash weight from DXA. The results of the study show that SOS is able to predict total BMD with adequate accuracy. If femoral BNID could be obtained in vivo with comparable accuracy, femoral QUS would be suited for the assessment of bone status at one of the main osteoporotic fracture sites. (c) 2006 Elsevier Inc. All rights reserved.