Jeremy F. Magland1, Chamith Rajapakse1, X H. Zhang2, X E. Guo2, Felix W. Wehrli1
1Department of Radiology, University of Pennsylvania Medical Center, Philadelphia, PA, USA; 2Department of Biomedical Engineering, Columbia University, New York, NY, USA
While structural trabecular bone parameters combined with bone volume fraction (BV/TV) can be used to predict bone strength beyond that predicted by BV/TV alone, a more direct assessment of strength on the basis of MR images may be found using micro-finite element (FE) modeling. A custom FE solver, specifically optimized for this application, was implemented in pure C++ and used to estimate mechanical bone parameters on the basis of grayscale images at in-vivo resolution. The algorithm was applied to -MR images as well as to -CT images downsampled to various resolutions. The results demonstrate the value of retaining grayscale information in the FE computation.