A dedicated computed tomography system was used to acquire transaxial images of the distal radius to assess trabecular bone structure in vivo. We segmented trabecular bone from the marrow and soft tissue background by postprocessing the image with a region grow and skeletonization step. From the processed image we assessed the integrity of the bone by examining the continuity of its trabecular network and by determining the area of the holes comprising its marrow space. The continuity of the bone imaged was assessed by a proposed connectivity index (CI) and the size of the marrow spaces was assessed by calculating a mean hole area (H_A) in the bone cross-section. Repeat measurements revealed that the intra-subject variability in CI and H_A was small (CV < 6%). Both CI and H_A were sensitive enough to reflect differences in structure at the head of the radius and at several sites along its shaft.

We tested the diagnostic value of assessing bone structure at the distal end of the radius by measuring trabecular bone density, CI and H_A in a mixed group of 26 subjects, nine of whom had suffered a wrist fracture. We found that a trabecular bone density threshold of 116 mg cm^-3, corresponding to two standard deviations below the mean density in the 17 non-fractured subjects, separated fractured from non-fractured subjects with a sensitivity of 22% and a specificity of 100%. A CI threshold of -4.7 doubled the sensitivity (44%) and maintained the 100% specificity. An H_A threshold of 4.5 mm² achieved a sensitivity of 89% and a specificity of 94%. This increased sensitivity achieved by our indices suggests that an in vivo assessment of trabecular bone structure can contribute significantly to the identification of persons at risk of fracture.