A Novel Approach to Fracture-Risk-Assessment in Osteoporosis by ROI-Oriented Application of the Minkowski-Functionals to Dual X-Ray Absorptiometry Scans of the Hip

摘要

Fractures of the proximal femur represent the worst complication in osteoporosis with a mortality rate of up to 50% during the first post-traumatic year. Bone mineral density (BMD) as obtained from dual energy x-ray absorptiometry (DXA) is a good predictor of fracture risk. However, there is a considerable overlap in the BMD-results between individuals who have fractured and those who have not. As DXA uses highly standardized radiographic projection images to obtain the densitometric information, it can be postulated that these images contain much more information than just mineral density. Lately, geometric dimensions, e.g. hip axis length (HAL) or femoral neck axis length (FNAL), are considered in conjunction with BMD, which may allow to enhance the predictive potential of bone mass measurements. In recent studies we sucessfully introduced a novel methodology for topological analysis of multi-dimensional graylevel datasets, that, for instance, allows to predict the ultimate mechanical strength of femoral bone specimens. The new topolocial parameters are based on the so called Minkowski Functionals (MF), which represent a set of topographical descriptors that can be used universally. Since the DXA-images are multi-graylevel datasets in 2D obtained in a standardized way, they are ideally suited to be processed by the new method. In this study we introduce a novel algorithm to evaluate DXA-scans of the proximal femur using quantitative image analysis procedures based on the MF in 2D. The analysis is conducted in four defined regions of interest in analogy to the standard densitometric evaluation. The objective is to provide a tool to identifiy individuals with critically reduced mechanical competence of the hip. The result of the new method is compared with the evaluation bone mineral density obtained by DXA, which - at present - is the clinical standard of reference.