Skip to content
Open access

Generalizable virtual staining from histopathology to spatial proteomics across biological scales

Nov 2026 · bioRxiv · 0 citations · 50 references
Biology

TL;DR

It is shown that routine hematoxylin and eosin (H&E) histology can provide a queryable, morphology-based representation of spatial protein expression that remains informative from pixels to patients.

Abstract

Large histology archives capture morphological correlates of molecular tissue states but lack direct spatial protein measurements. In this study, we show that routine hematoxylin and eosin (H&E) histology can provide a queryable, morphology-based representation of spatial protein expression that remains informative from pixels to patients. To achieve this, we developed QUEST, a model framework that represents protein markers as semantic queries for flexible multiplex immunofluorescence (MIF) prediction. Trained on a heterogeneous atlas of paired H&E and MIF images comprising over 10,000 tissue regions, QUEST captured morphological patterns associated with protein markers and reconstructed their expression intensities across diverse tissue types and cohorts. Its utility extended beyond image reconstruction and preserved biologically meaningful information across multiple scales: virtual staining preserved protein marker expression at the cellular and tissue scales, supported cell phenotyping, cellular neighborhood analysis, and characterization of spatial tissue organization, retained clinically relevant signals for patient stratification, and enabled H&E-based queries to retrieve MIF references with similar molecular profiles. These findings position virtual staining not simply as image generation but as a flexible interface between routine histology and spatial proteomics, expanding the analytical value of existing H&E archives.

Read PDF

Similar papers

Review Open access Sep 2026

Integrating AI-Based Digital Pathology with Spatial Transcriptomics for Deeper Insight.

It is argued that the most credible translational goal is not to infer every transcript from H&E, but to identify biologically coherent spatial states with reproducible morphological correlates that can be validated across independent platforms and cohorts before being mapped across large archival collections.

M. A. Rezania, Haechan Yi, Jae-Il Choi et al. · 0 citations
Oct 2026

When an age-associated stain becomes a spatial phenotype.

Histochemical staining patterns may vary with age because of changes in cell number, differentiation state, marker expression, tissue geometry, or pre-analytical conditions. Recent data indicate that epithelial tuft-cell abundance decreases and LGR5 immunoreactivity changes in the ageing human colon. We emphasize that...

M. Vijayasimha, Sonika Yadav, Bhawana Thukral · 0 citations
Review Open access 2026

Predicting spatial transcriptomics from histology images: Progress, challenges, and directions

Spatial transcriptomics (ST) maps gene expression within intact tissues, but its cost and technical complexity limit broad adoption. A growing body of deep-learning methods now seeks to infer ST directly from routine histology images, especially hematoxylin and eosin (H&E)-stained slides, with the aim of turning archiv...

Ke-Xin Xu, H. Koohy, A. Antanaviciute · 0 citations

MoSPR: Histology-to-Gene Expression Prediction with Morpho-Spatial Macrostates and Low-Rank Molecular Programs

MoSPR (Morpho-Spatial Program Regression), a linear framework that couples an adjacency-informed histology representation with a low-rank molecular basis, provides an interpretable link between spatially coherent macrostate regions and their associated molecular programs.

Dongmyung Shin, Geong-Ran Lee, Yesung Cho et al. · 0 citations
Preprint Nov 2026

Predicting Single-Cell Spatial Gene Expression from H&E using Pathology Foundation Models

Predicting gene expression directly from hematoxylin and eosin (H&E) images offers a scalable and cost-effective approach to deriving spatial molecular insights from routine histology. Pathology foundation models, typically pretrained using tile-level objectives, may not adequately capture the fine-grained cellular fea...

Ashley P. Tsang, Evan Liu, Conrad Foo et al. · 0 citations
Open access Oct 2026

Fine-Grained and Reliable Cell Classification from Pathology Images

CellBridge is introduced, a multi-modal contrastive learning framework pretrained on 16M cells with paired H&E and spatial profiles, for transferring molecular information into morphological representations that consistently achieves smaller prediction sets, while reducing class-wise coverage gaps.

Sevda Öğüt, Arpit Merchant, Amirhossein Vahidi et al. · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.