Evaluation of direct strain field prediction in bone with data-driven image mechanics (D2IM-Strain)

This work proposes D2IM-Strain, an upgrade to the data-driven image mechanics framework that predicts strain fields directly from undeformed X-ray computed tomography images of bone. The direct strain prediction model significantly improved accuracy for strain magnitudes below 10000 microstrain and reduced false-positive high-strain classifications by 75%, representing a critical step toward robust data-driven volume correlation methods for hierarchical materials.

Machine learning (ML)
Digital volume correlation (DVC)
Bone mechanics
X-ray tomography (XCT)
Strain prediction
A data-driven image mechanics approach (D2IM-Strain) that predicts strain fields directly from XCT images of bone, eliminating noise from numerical differentiation of displacement fields.
Authors

Jon Valijonov

Peter Soar

James Le Houx

Gianluca Tozzi

Published

March 31, 2026

Preprint
Back to top