VERTEBRAL STRAIN MEASUREMENT: A COMPARISON OF IMAGE REGISTRATION VS. FINITE ELEMENT ANALYSIS

Abstract

To compare strains measured in a whole rat-tail vertebra by image registration (IM) with those predicted by solid finite element analysis (FEA). Quantification of bone strain allows better understand fracture risk, bone healing and turnover.

The sixth caudal vertebra of an rnu/rnu rat was μCT scanned (17.5×17.5×17.5μm/voxel) while loaded (27N axial compression) and unloaded. IM was used to calculate strain and displacement fields in the bone due to the applied load by finding a spatial mapping between the two scans. Strain was resolved to varying spatial resolution; high strain gradient regions (ie growth plates) were analyzed to higher spatial resolutions. A FE model was created of the unloaded vertebra, consisting of tetrahedral elements with transversely isotropic material properties. Elements were assigned elastic moduli based upon μCT image intensities. Growth plate moduli ranged from 0–150kPa and the bone moduli ranged from 0.2–15000MPa. Vertebral geometry was created through segmentation of μCT images. Displacement boundary conditions were obtained by matching cranial and caudal surfaces in the unloaded and loaded scans. The displacement fields of the two methods were compared by using the fields calculated to deform the unloaded scan to match the loaded scan. The strains were compared by plotting FEA measured axial strain against IM calculated axial strain.

The displacement fields calculated by both methods were able to spatially align the unloaded scan to the loaded scan (Mean Voxel Intensity Difference: FEA=441HU, IM=328HU, Unregistered=969HU). IM and FEA show very limited agreement in axial strain measurement (R2=0.388, Slope=0.75, X-Intercept=0.0037) although both calculated high axial strains in the growth plates and low axial strains in the trabecular and cortical bone. Good agreement was found in the mean axial strain measured by both methods (IM= −0.044, FEA=−0.037). IM was better able to deal with difficulties in quantifying bone strain due to the growth plate than FEA.

IM presents advantages over FEA in measuring strain in complex whole bone trabecular structures, however has lower spatial resolution than is possible with FEA.