Use of Digital Image Correlation for Validation of Surface Strain in Specimen-specific Vertebral Finite Element Models

Finite element (FE) models of the spine are typically validated to be similar to the average experimental response of specimens with anatomical variation, such as varying sizes and degrees of degeneration. To account for this variation in a manner that allows validation against a specific vertebral response instead of an average range, specimen-specific models can be developed. The purpose of this study was to compare the strain predicted by specimenspecific FE models of vertebrae under compressive loading with the strain measured experimentally using digital image correlation (DIC), a non-contact optical method. A secondary purpose was to compare the DIC measured strain with strain gauges to verify the use of DIC for obtaining full field strain. DIC has not been used to investigate cadaveric vertebrae previously. Fresh-frozen human cadaveric vertebrae (n = 6) from three donors were obtained. A high resolution CT scan of each bone was taken and the bone was segmented from the CT images to create geometric models for the FE analysis. For the experiment, the vertebrae were tested in a materials testing machine. The bone was loaded dynamically to failure at a rate of 0.5 m/s. For the FE model, heterogeneous Young’s modulii were assigned to each of the trabecular elements. A displacement function was applied at the superior endplate based on the experimental measurements. Before the loading, the average error between the DIC measured strain and strain gauge was 105.2 microstrain while during the loading it was 417.3 microstrain. We judged this error in the DIC acceptable to use for model validation. The minimum principal strains from the experiment and FE model were compared qualitatively and quantitatively. The coefficients of determination for the quantitative comparisons were between 0.15 and 0.41. The similarity shows the potential of the specimen-specific models to predict surface strain, but further refinement of the models is needed. Use of DIC provides a unique and detailed dataset for validation of specimen-specific models.