The Quantitative Bone Microarchitectural Based Computer Aided Analysis for Osteoporosis Detection Using Digital Hip Radiograph

The quality of life can be improved and risk of fractures would be effectively reduced, by the proper initial diagnosis of the osteoporotic abnormality. The identification of low bone mineral density (BMD) or the onset of osteoporosis can be performed by means of detecting the extracted trabecular features from digital hip radiographs. The objective of our study was to apply a gradient based image processing technique from simple X-ray in order to measure trabecular structure for osteoporosis evaluation and its associated fracture risks. Fifty South Indian women were chosen for whom DXA (DPX Prodigy, GE-Lunar Corp., USA) as well X-ray (Multiphos, Siemens, Germany) measurements were acquired from the right proximal region. For the digital X-rays acquired, gradient based texture image analysis was performed; multiple linear regression models which is the combination of age and gradient texture has outputted predicted BMD in order to distinguish normal group from osteoporotic one. None of the textural (run-length) variables have exhibited significant correlation with DXA variables in case of original radiographs. Whereas, the gradient based method has demonstrated significant co-relational emphasis with textural (HGRE High gray-level run emphasis) parameter with Neck and Ward’s region of DXA BMDs. Also, the regression model has displayed significant co-relational impact with all anatomical sites of DXA BMDs. Our results emphasize that the predicted BMD extracted from digital hip X-ray image would be useful as a low cost screening tool for osteoporosis diagnosis and its associated fracture risks.