sCellST predicts single-cell gene expression from H& E images

TL;DR

This paper introduces a deep learning model that predicts single-cell gene expression from H&E images, enabling detailed molecular insights from standard histological slides.

Contribution

The novel approach uses a Multiple Instance Learning framework to infer single-cell gene expression from full-slide morphology, improving resolution and biological relevance.

Findings

The model matches patch-based methods in spot-level prediction tasks while capturing fine-grained morphological variation.

It recovers biologically meaningful gene expression patterns across two cancer datasets.

The approach distinguishes fine cell populations and enables molecular-level interpretation of histological slides at scale.

Abstract

Understanding the spatial organization of individual cell types within tissue and how this organization is disrupted in disease, is a central question in biology and medicine. Hematoxylin and eosin-stained slides are widely available and provide detailed morphological context, while spatial gene expression profiling offers complementary molecular insights, though it remains costly and limited in accessibility. Predicting gene expression directly from histological images is therefore an attractive goal. However, existing approaches typically rely on small image patches, limiting resolution and the ability to capture fine-grained morphological variation. Here, we introduce a deep learning approach that predicts single-cell gene expression from morphology, matching patch-based methods on spot level prediction tasks. The model recovers biologically meaningful expression patterns across two cancer datasets and distinguishes fine cell populations. This approach enables molecular-level interpretation of standard histological slides at scale, offering new opportunities to study tissue organization and cellular diversity in health and disease.

Predicting gene expression from H&E slides is a cost-effective alternative to Spatial Transcriptomics for clinical use. Here the authors introduce a Multiple Instance Learning approach to infer single-cell expression, unlike prior methods which operate at coarser patch level.

Figures