<title>Optical and digital processing of radiographs for the early detection of osteoporosis</title>

Abstract

The diagnosis of osteoporosis has traditionally been made by observing bone radiographs and analyzing the optical density. More advanced methods measure the mineral content or the quantity of bone tissue, but do not determine bone quality, which depends on the bone tissue architecture. The structure and organization of bone tissue is a vital factor for assessing bone strength, as minute changes in the structure strongly affect bone strength. This study describes a novel method, based on optical and digital image processing of conventional radiographs for a quantitative evaluation of bone tissue architecture. The trabecular pattern appearing on the femoral neck radiograph was analyzed based on the power spectrum of this pattern, which was obtained by optical Fourier transform. The two-dimensional optical power spectrum was captured to the computer by a CCD device and after filtering and noise reduction, it was characterized using spatial frequencies in different directions. Software was developed for the analysis of the power spectrum of the trabecular pattern to yield a bone index. The effect of the image optical density on the calculated bone index was not significant. On the other hand, the bone index, based on the trabecular structure, was found to differ significantly between normal and osteoporotic bones. Thus, the power spectrum analysis could differentiate normal from osteoporotic bones much better than the amount of radiation absorbed by the bone.