Image-guided multiscale modeling of trabecular bone: lacuna-resolved stress amplification and interaction effects

Osteocyte lacunae are ubiquitous microstructural cavities in bone that perturb local stress and strain fields, shaping the mechanical microenvironment relevant to mechanotransduction and microdamage initiation. Yet lacunar-scale mechanical fields are not directly measurable in situ, and existing numerical studies are often limited by idealized geometries, incomplete verification, generic material properties, and a lack of specimen-matched experimental validation. Here, we present an image-guided multiscale framework integrating synchrotron radiation µCT in situ compression, convolutional neural network (CNN)-based lacunar segmentation and morphometrics, inverse identification of tissue elastic properties, and µCT-driven finite element simulation with lacuna-resolved submodeling. In a human trabecular bone case study, CNN-based analysis identified 219,931 lacunae and equivalent-ellipsoid reconstruction reproduced segmented lacunar volumes with high fidelity (R² = 0.97), enabling robust population-scale shape and orientation assessment and revealing preferential alignment with the local trabecular axis. Numerical verification supported a loose-wrap hexahedral discretization and mesh convergence at h = 0.04 mm (relative discretization error of 0.35%). Specimen-specific inverse calibration reproduced the experimental linear force-displacement response (RMSE = 0.82 N). Lacuna-resolved analyses across three submodels showed localized equatorial tensile amplification (K_t=1.38-1.99) governed primarily by lacunar elongation, orientation, and local stress triaxiality. These effects were captured by a compact multivariate power-law relation (R²=0.81; R_LOOCV²=0.72). Inter-lacunar interactions decayed rapidly with distance and became negligible beyond ∼8-9 effective radii (∼23 µm), providing a scale-free isolation criterion. This experimentally calibrated and numerically verified workflow bridges specimen-scale mechanics and peri-lacunar stress fields and enables quantitative, population-scale assessment of how lacunae modulate local mechanical stimuli. STATEMENT OF SIGNIFICANCE: Osteocytes sense mechanical loading through microscopic cavities (lacunae) embedded in bone, yet the local mechanical environment around these cavities cannot be measured directly and is often oversimplified in computational studies. This work introduces a validated, image-guided multiscale framework that combines synchrotron imaging, deep-learning-based lacunar mapping, in situ mechanical testing, and specimen-calibrated finite element modeling to quantify how real lacunar shape, orientation, and spacing affect local stress amplification. Unlike previous idealized or unvalidated approaches, the method provides transferable, quantitative descriptors of lacuna-driven mechanical perturbations and defines a scale-free criterion for lacunar interaction. The framework enables population-scale assessment of osteocyte-level mechanics and offers new tools for studying bone quality, adaptation, and fragility beyond bone mass alone.