Predicting Single-Cell Spatial Gene Expression from H&E using Pathology Foundation Models
Organizations: AI for Biology & Translation, Computational Sciences Center of Excellence, Genentech · Gilbert S. Omenn Department of Computational Medicine & Bioinformatics, University of Michigan · Computational Biology & Medicine, Computational Sciences Center of Excellence, Genentech · Digital and Spatial Pathology, Translational Genomics and Proteomics, gRED, Genentech
Abstract
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 features required for single-cell molecular prediction. In this work, we systematically evaluate and establish domain-informed and task-specific strategies for adapting pathology foundation models to single-cell spatial gene expression prediction. Using paired H&E and Xenium 5K data, our approach improves mean gene-wise Pearson correlation by 2.67x over the standard zero-shot baseline and achieves state-of-the-art performance on this task. We demonstrate strong gene-detection performance across the full gene panel (AUROC = 0.800) and importantly identify that predictive performance varies substantially across biological contexts, with our model achieving its highest Pearson correlation for spatially variable genes (r = 0.643). Our findings demonstrate that, despite the challenges of sparse single-cell measurements, H&E-derived representations can recover biologically meaningful spatial expression patterns, highlighting both the opportunities and limitations of morphology-based molecular inference at cellular resolution.
Figures & tables
| Model | SVG50 | IBD | HVG50 | 5K |
|---|---|---|---|---|
| OmiCLIP Chen et al. (2025) | 0.207 | 0.072 | 0.057 | 0.029 |
| BLEEP Xie et al. (2023) | 0.349 | 0.160 | 0.117 | 0.065 |
| GHIST Fu et al. (2025) | 0.505 | 0.167 | 0.087 | 0.081 |
| SpatialEx Liu et al. (2026) | 0.509 | 0.207 | 0.158 | 0.094 |
| Baseline † | 0.469 | 0.182 | 0.151 | 0.065 |
| + MLP | 0.525 ( ) | 0.235 ( ) | 0.195 ( ) | 0.110 ( ) |
| Model | Native granularity | PCC (per-gene) | PCC (per-cell) | AUROC |
|---|---|---|---|---|
| OmiCLIP Chen et al. (2025) | Spot-level | 0.029 0.006 | 0.203 0.026 | 0.568 0.016 |
| BLEEP Xie et al. (2023) | Spot-level | 0.065 0.041 | 0.124 0.078 | 0.578 0.062 |
| GHIST Fu et al. (2025) | Single-cell | 0.081 0.007 | 0.242 0.027 | 0.605 0.012 |
| SpatialEx Liu et al. (2026) | Single-cell | 0.094 0.009 | 0.242 0.031 | 0.642 0.005 |
| Ours (frozen) | Single-cell | 0.161 0.016 | 0.271 0.034 | 0.786 0.005 |
| Ours (LoRA) | Single-cell | 0.174 0.014 | 0.280 0.033 | 0.800 0.009 |
Appendix figures & tables8 assets
Supplementary material from the paper’s appendix.
Appendix
| Configuration | PCC (5K) | |
|---|---|---|
| Cell mask (reference) | 0.161 0.016 | — |
| Disc µm, Gaussian µm | 0.154 0.016 | 0.007 ( 4.6%) |
| Disc µm, unweighted mean | 0.149 0.015 | 0.012 ( 7.3%) |
| Class-token (baseline) | 0.128 0.013 | 0.035 ( 20.2%) |
| Input | Pixel size | FOV | Token width | Tokens/cell | PCC (5K) |
|---|---|---|---|---|---|
| ( m/px) | ( m) | ( m) | median (mean) | ||
| 0.2125 | 47.6 | 2.98 | 4 (5.1) | 0.1611 0.0158 — | |
| 0.5 | 112 | 7.00 | 1 (1.2) | 0.1521 0.0125 5.6% | |
| 1.0 | 224 | 14.0 | 1 (1.0) | 0.1292 0.0107 19.8% | |
| Single-cell crop | 0.2125 | 14.0 | 0.88 | 50 (56.8) | 0.1361 0.0186 15.5% |
Spatial Gene Expression Prediction
Spatial gene expression prediction infers tissue-resolved molecular activity from histology and spatial context.
See all 47 papers