Dirk Mueller MD, Presenter: Nothing to Disclose

Roberto Adrian Monetti PhD, Abstract Co-Author: Nothing to Disclose

Sharmila Majumdar PhD, Abstract Co-Author: Nothing to Disclose

Christoph W. Raeth PhD, Abstract Co-Author: Nothing to Disclose

Ernst Josef Rummeny MD, Abstract Co-Author: Nothing to Disclose

Thomas Marc Link MD, Abstract Co-Author: Nothing to Disclose

In this study the scaling vector method (SVM) was applied to analyze the trabecular bone structure of high resolution magnetic resonance (HR-MR) images of the calcaneus. Using the SVM a 3D non-linear texture measure which takes into account the anisotropic nature of the trabeculae was extracted. This new structure analysis technique was compared with BMD in its diagnostic performance to differentiate postmenopausal patients with and without osteoporotic spine fractures.

HR-MR imaging of the calcaneus was performed at 1.5 T in 85 women (45 postmenopausal women with osteoporotic spine fractures and 40 postmenopausal controls) using a custom built receive-only phased-array coil. We used a three-dimensional gradient-echo sequence with an axial orientation, a slice thickness of 0.5 mm, a TE of 5.7 ms, a TR of 28.1 ms, a flip angle of 30° and in plane spatial resolution of 195x195 µm². After normalization of the images and selection of the region of interest (ROI) structure analysis was performed using algorithms based on the new anisotropic SVM, which is respecting the orientation of the trabeculae of the analyzed bone. In addition BMD measurements of the spine, calcaneus and proximal femur using dual energy X-ray absorptiometry (DXA) were obtained in all patients.

Significant differences between both patient groups were obtained using structure analysis and BMD (p<0.01). Receiver operating characteristics (ROC) analyses were used to determine the diagnostic performance in differentiating both groups. In comparison with BMD of the spine (Area under curve (AUC) = 0.75), BMD of the calcaneus (AUC = 0.71) and BMD of the proximal femur (AUC = 0.84) the best results were found for the 3D scaling vector method (AUC = 0.88).

The results of this study show that the new anisotropic algorithm based on a local 3D SVM which is respecting the orientation of the trabeculae may improve the diagnostic performance in differentiating postmenopausal women with and without osteoporotic spine fractures.

Mueller, D, Monetti, R, Majumdar, S, Raeth, C, Rummeny, E, Link, T, The Anisotropic 3D Scaling Vector Method to Investigate Trabecular Bone Structure in High Resolution Magnetic Resonance Imaging of the Calcaneus.  Radiological Society of North America 2004 Scientific Assembly and Annual Meeting, November 28 - December 3, 2004 ,Chicago IL. http://archive.rsna.org/2004/4411929.html