The abstract is read at the publisher; the summary is JClub's.
CarHE predicts high-dimensional spatial transcriptomic profiles from routine H&E slides with high accuracy across datasets and spatial transcriptomics platforms, including TLS-associated regions.
- Why it matters: Spatial transcriptomics is costly and limited in resolution, scalability, and gene coverage, restricting large-scale studies and broader adoption.
- What they did: CarHE uses multimodal contrastive learning to align cell type–specific transcriptomic signals with histological features from H&E images, then infers spatially organized transcriptomic context.
- The result: In a cohort of 880 lung cancer patients, CarHE-derived features are associated with disease-free survival and outperformed current approaches, enabling scalable H&E-based spatial inference.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.