Quantitative assessment of cortical bone remodelling from routine radiographs of total hip arthroplasty

In orthopaedic clinical studies, quantitative X-ray bone densitometric analysis could provide precise means of detecting alterations within the bone. With duel-energy X-ray absorptiometry and diphotonic absorptiometry precise and reproducible measures are possible, but they require a specific and costly examination. The use of routine and past available radiographs could facilitate long-term follow-up retrospective studies, but when femoral X-ray bone density is assessed by the naked eye on conventional radiographs, the percentage of errors may be very high (80%). The advantage of a method using conventional radiographs has led us to look for an original technology to estimate X-ray bone density from routine radiographs using both soft tissues and prostheses as X-ray absorptiometric phantoms. In theory, this calculation requires several approximations. An experimental study based on 40 subjects aimed at determining the rate of error due to these approximations. Three radiographs were made for every subject with very different exposure settings. They were computerized by a CCD camera to measure the grey levels (256 scale) of soft tissue and prosthesis, as well as variations in contrast. The statistical analysis showed that when comparing X-ray bone density on different routine radiographs, the percentage of error could be lower than 10% but only under these conditions: (a) grey levels of soft tissue > 26; (b) grey levels of prosthesis < 212; (c) variation of contrast between radiographs > 0.79; (d) radiographs made with the same incidence, and (e) weight of the subject remaining constant during the whole period. It therefore seems possible to compare retrospectively the cortical X-ray bone density from routine radiographs, and to quantify bone remodelling around total hip arthroplasty, but only for the same subject.