Model-Free Multiparametric Analysis of SHG Images Reveals Collagen Signatures in Breast Cancer

We present a novel multiparametric approach for analyzing high-resolution polarization-resolved Second-Harmonic Generation (pSHG) images to identify aligned collagen bundles in histological samples of human breast cancer. This method introduces a model-free optical biomarker—Polarization Asymmetry (PA)—which quantify collagen supramolecular organization without relying on assumed symmetry models. This biomarker was compared with both the angular phase shift (ϕ ₀ ) of pSHG curves and the |χ _yyy | component of the second-order susceptibility tensor, derived from fitting pSHG data to a trigonal symmetry model. Our results reveal that high values of PA, |χ _yyy |, and ϕ ₀ variation over micrometric scale, consistently detect the presence of aligned collagen bundles, suggesting a correlation with the structural remodeling of the extracellular matrix at the sub-micrometric scale that may influence tissue mechanical properties. Validation on human corneal tissue, characterized by well-known collagen architecture and biomechanics, confirmed the robustness and generalizability of our approach for detecting diagnostically relevant collagen signatures in cancer and other collagen-rich tissues.