Statistical analysis of bone mineral density using voxel-based morphometry-an application on proximal sesamoid bones in racehorses.

Abstract

Voxel-based morphometry (VBM) is a computational technique that has been used to analyze statistical differences between groups of MR brain images. This study outlines a new VBM pipeline, designed for determining statistical variation in bone mineral density (BMD). CT images of proximal sesamoid bone (PSB) specimens from the right forelimb of six racehorses that had suffered PSB fractures were compared with six age-matched control specimens. Following segmentation, masked gray-scale images were co-aligned to a statistical template generated with all 12 CT datasets iteratively. Student t-tests were performed voxel-by-voxel on spatially aligned 3D images to reveal significant differences in the spatial variation of bone density between the fracture and control groups. Overall density and densities from the axial and abaxial areas of PSBs were compared between groups. The BMD in abaxial regions of the medial and lateral PSBs of the fracture group were 12.7% (p = 0.044) and 13.5% (p = 0.047) higher, respectively, than controls. The overall mean density of paired PSB and the medial and lateral PSBs separately were higher in the fracture group. The VBM pipeline facilitates detailed comparison of density variation between bone groups at the voxel level.