Computational Mueller matrix microscopic pathology based on differential decomposition for medical reference mapping

Introduction Label-free quantitative polarimetric imaging provides access to physically interpretable tissue properties that are not available from conventional intensity-based microscopy, offering new opportunities for computational digital pathology and tissue bioinformation analysis. By characterizing the interaction between polarized light and tissue microstructure, Mueller matrix imaging enables spatially resolved quantification of optical mechanisms including diattenuation, birefringence, and depolarization. However, translating this information toward tissue-scale pathological assessment remains challenging because high spatial resolution is inherently associated with a limited field of view, while the resulting high-dimensional polarimetric information is difficult to interpret directly and relate to conventional medical references. Methods Here, we present a physics-informed computational framework, termed Computational Mueller Matrix Microscopic Pathology, based on differential Mueller matrix decomposition. The framework integrates differential decomposition, wide-field Mueller matrix image stitching, and mechanism-specific multi-channel pseudo-color visualization into a unified computational pipeline. By combining physically interpretable polarimetric observables with a SIFT-based stitching strategy, the framework extends high-resolution Mueller matrix imaging to large tissue areas while preserving the spatial consistency and physical interpretability of the reconstructed polarimetric information. Mechanism-specific pseudo-color mappings further transform the decomposed optical responses into spatially distinguishable representations, facilitating their visual and quantitative comparison with conventional histological references. Results The framework was evaluated on four paired H&E-stained murine melanoma specimens under untreated (Trt − ) and treated (Trt + ) conditions through complementary quantitative, physical, and histological analyses. For these thin pathological sections, differential decomposition approaches and Lu–Chipman decomposition yielded comparable tissue representations, while mechanism-specific pseudo-color mappings enabled the resulting polarimetric contrasts to be visualized according to their underlying optical mechanisms. Notably, depolarization consistently emerged as a tumor-associated contrast across the polarimetric numerical distributions, spatial heterogeneity, and visual enhancement of differential pseudo-color mappings, with preliminary spatial correspondence to H&E-defined tissue regions. These observations indicate that physically interpretable polarimetric contrasts can capture spatially resolved information associated with tissue structural characteristics. Discussion Overall, the results demonstrate the feasibility of combining differential Mueller matrix decomposition, large-field polarimetric imaging, and mechanism-specific pseudo-color visualization for tissue-scale pathological assessment. Beyond enhanced visualization, the proposed framework establishes quantitative links between histological appearance and physically interpretable optical properties, providing a foundation for objective tissue characterization and the extraction of biologically relevant spatial information. These findings suggest that Mueller matrix polarimetry, combined with high-magnification large-field imaging and differential decomposition, can provide spatially resolved polarimetric features relevant to the characterization of melanoma-associated tissue alterations and treatment-related structural changes.