The abstract is read at the publisher; the summary is JClub's.
sc-HistoMorphomics profiles nuclear morphology from whole-slide H&E and links it to spatial transcriptomics using 194 interpretable features across endometrium, breast cancer, and pancreas.
- Why it matters: Quantitative nuclear morphology is informative but has been difficult to measure at scale and connect to spatial molecular state in tissues.
- What they did: The framework uses deep-learning segmentation on H&E images, then extracts 194 morphometric features spanning nuclear shape, eosin/hematoxylin texture, and neighborhood architecture, and correlates them with spatial transcriptomics.
- The result: It finds conserved feature structure across tissues and reproducible morphology–gene expression associations, including eosin texture linked to senescence in endometrium, morphology clustering matching stromal/lymphocyte/epithelial tumor classes in breast cancer, and CDKN1A negative correlations with multiple morphological features in pancreas across cell types and donors.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.