Digital tomosynthesis based digital volume correlation: A clinically viable noninvasive method for direct measurement of intravertebral displacements using images of the human spine under physiological load.

Abstract

We have developed a clinically viable method for measurement of direct, patient-specific intravertebral displacements using a novel digital tomosynthesis based digital volume correlation technique. These displacements may be used to calculate vertebral stiffness under loads induced by a patient's body weight; this is particularly significant because, among biomechanical variables, stiffness is the strongest correlate of bone strength. In this proof of concept study, we assessed the feasibility of the method through a preliminary evaluation of the accuracy and precision of the method, identification of a range of physiological load levels for which displacements are measurable, assessment of the relationship of measured displacements with microcomputed tomography based standards, and demonstration of the in vivo application of the technique. Five cadaveric T11 vertebrae were allocated to three groups in order to study (a) the optimization of digital volume correlation algorithm input parameters, (b) accuracy and precision of the method and the ability to measure displacements at a range of physiological load levels, and (c) the correlation between displacements measured using tomosynthesis based digital volume correlation vs. high resolution microcomputed tomography based digital volume correlation and large scale finite element models. Tomosynthesis images of one patient (Female, 60yr old) were used to calculate displacement maps, and in turn stiffness, using images acquired in both standing and standing-with-weight (8kg) configurations. We found that displacements were accurate (2.28µm total error) and measurable at physiological load levels (above 267N) with a linear response to applied load. Calculated stiffness among three tested vertebral bodies was within an acceptable range relative to reported values for vertebral stiffness (5651-13260N/mm). Displacements were in good qualitative and quantitative agreement with both microcomputed tomography based finite element (r2 =0.762, P<0.001) and digital volume correlation (r2 =0.799, P<0.001) solutions. For one patient tested twice, once standing and once holding weights, results demonstrated excellent qualitative reproducibility of displacement distributions with superior endplate displacements increasing by 22% with added weight. The results of this work collectively suggest the feasibility of the method for in vivo measurement of intravertebral displacements and stiffness in humans. These findings suggest that digital volume correlation using digital tomosynthesis imaging may be useful in understanding the mechanical response of bone to disease and may further enhance our ability to assess fracture risk and treatment efficacy for the spine.