Introduction: Osteoporosis is diagnosed by measurement of bone mineral density (BMD), with techniques that include single and dual energy x-ray absorptiometry (SXA, DXA) and quantitative computed tomography (QCT), which have relatively high cost and limited availability. Radiographic absorptiometry (RA), the corresponding technique based on plain radiographs, has not been favoured clinically because it required complex devices to calibrate and difficult techniques to measure the optic densitometry of films. In this paper a newly developed computerised radiographic technique is described that gives results in actual BMD units and includes a correction for soft-tissues. The technique has been developed in conjunction with a new solid phantom to facilitate the use in fracture clinics, without the need for further patient exposure. It also has the advantage of measuring the ultradistal radius, a clinically useful cancellous rich site. The design of the phantom and initial calibration testing are described.

M& M: The new phantom is constructed from a solid water equivalent material, removing the need for a soft-tissue equalising water bath. The necessary hydroxyapatite (HA) calibration wedge is built into the phantom to give results in actual BMD units. A high-density grid has been used to minimise the effects of scatter caused by soft tissues. The phantom was initially calibrated by comparing its results with those obtained from specially manufactured hydroxyapatite disks. The European Forearm Phantom (EFP) was used in studies of further calibration and testing of anatomical specimens, obtained with permission from the department of Anatomy at the University of St. Andrews.

Results: The test comparing the BMD determined by the new phantom to the known density of specially manufactured HA discs yielded a very high correlation (R> 0.999) with an error < 2%. Further studies with the EFP as a reference standard confirmed a very high correlation (R> 0.999) with an error of < 5%.

Finally a study was performed on 25 cadaver forearms that were available for imaging using DXA and the modified RA technique. BMD results showed good correlation (R=0.93, P< 0.001) with an error of less than 10%. The error is partly explained because during this part of the study a high-resolution anti-scatter grid was unavailable and a standard grid had to be used. Also some cadavers had previous fractures and dissection marks. Without these problems, it is expected that the error percentage would be significantly reduced.

Discussion: The phantom is light, small and the patient can comfortably fit even if the limb is fractured. The results compare favourably with published data for existing techniques that also used the EFP as a reference standard. The described technique carries the promise of a simple, inexpensive system, which yields superior results and can be used seamlessly in a fracture clinic. A pilot clinical study is being planned, comparing the results of the modified RA technique with peripheral DXA. This system can be quickly made available at minimum cost utilizing existing equipment. All the other advantages of computerised x-rays including more precise measurements of deformity and central analysis from a distance for isolated practices apply.