Classifying Bone Trabecular Volumes Using 3D Shape Descriptors

In this poster we apply the 3D-cubeness shape descriptor from [1] (C(S)) to a sample (n=37, vertebra, male) of cubic trabecular bone volumes obtained by microcomputerized tomography (CT). Also theoretical models with a decreasing number of assigned trabecular voxels, irrespective their position, were analyzed. Our results show that C(S) correctly classifies the theoretical and the trabecular bone volumes according to their bone volume density (BV/TV) parameter (r=0.97; r2=0.95), and trabecular fractal dimension (FD-BV) (r=0.87; r2=0.80). This provides an alternative to standard global 3D-morphometrical methods used for assessing the microarchitectural deterioration of trabecular bone.