Beyond H&E: Unlocking Pathological Insights with Polarization Imaging

TL;DR

The paper addresses the gap in histopathology where standard H&E misses birefringence and tissue anisotropy by introducing polarization imaging as a complementary modality. It builds a paired Polarization Pathology Dataset and proposes PolarHE, a self-supervised dual-modality fusion framework that decomposes representations into common and modality-specific components to fuse H&E with polarization information at the feature level. The authors achieve state-of-the-art patch classification on Chaoyang and MHIST, with accuracies of 86.70% and 89.06%, and show that polarization-guided pretraining improves the H&E encoder even when polarization data are unavailable at test time. The work demonstrates the potential of polarization imaging to provide interpretable optical cues and more robust, generalizable models for computational pathology.

Abstract

Histopathology image analysis is fundamental to digital pathology, with hematoxylin and eosin (H&E) staining as the gold standard for diagnostic and prognostic assessments. While H&E imaging effectively highlights cellular and tissue structures, it lacks sensitivity to birefringence and tissue anisotropy, which are crucial for assessing collagen organization, fiber alignment, and microstructural alterations--key indicators of tumor progression, fibrosis, and other pathological conditions. To bridge this gap, we construct a polarization imaging system and curate a new dataset of over 13,000 paired Polar-H&E images. Visualizations of polarization properties reveal distinctive optical signatures in pathological tissues, underscoring its diagnostic value. Building on this dataset, we propose PolarHE, a dual-modality fusion framework that integrates H&E with polarization imaging, leveraging the latter ability to enhance tissue characterization. Our approach employs a feature decomposition strategy to disentangle common and modality specific features, ensuring effective multimodal representation learning. Through comprehensive validation, our approach significantly outperforms previous methods, achieving an accuracy of 86.70% on the Chaoyang dataset and 89.06% on the MHIST dataset. These results demonstrate that polarization imaging is a powerful and underutilized modality in computational pathology, enriching feature representation and improving diagnostic accuracy. PolarHE establishes a promising direction for multimodal learning, paving the way for more interpretable and generalizable pathology models.

Figures (3)

• Figure 1: Polarization Imaging Acquisition and Processing System.
• Figure 2: Polarization Property (Fast-axis Orientation and Depolarization Saturation)
• Figure 3: Integrating H&E and polarization representations via feature decomposition.