Image-informed biphasic constitutive modeling reveals degeneration-dependent evolution of composite disc material behavior

Using MRI data from 30 volunteers spanning four degeneration stages, we developed an image-informed constitutive modeling framework that integrated deep learning-based tissue delineation with biphasic-swelling simulations. The segmentation framework achieved a mean Dice coefficient of 0.93 ± 0.02. Radiomic descriptors were linked with stage-dependent material properties, including fixed charge density and hydraulic permeability, and the resulting mechanical responses were evaluated under free-swelling, creep, and stress-relaxation. Radiomic–mechanical associations were stage dependent, with the strongest statistically supported association observed in healthy discs between major axis length and equilibrium residual force ( ρ = 0.85, 95% CI 0.67–0.95, q < 0.001). Equilibration time ( τ _95%) peaked in the moderate degeneration group across all three loading modes, increasing from 2.134 ± 0.490 h to 6.996 ± 1.939 h in free-swelling, from 8.018 ± 1.011 h to 15.105 ± 3.142 h in creep, and from 0.428 ± 0.090 h to 1.085 ± 0.641 h in stress-relaxation, suggesting a transport-limited transitional stage. This research demonstrates a reproducible workflow for linking non-invasive imaging with constitutive modeling to fundamental material laws, providing a quantitative basis for investigating stage-dependent changes in disc mechanics.