Background: Predicting genomic alterations from routine hematoxylin and eosin (H at the patch-level, prediction of APCSNV reached an AUROC of 0. 916, and prediction of KRASSNV reached an AUROC of 0. 811 on the held-out test set. Conclusions: In a heterogeneous clinical gene-panel setting, pathology foundation models can provide strong baseline genomic-prediction signals without additional fine-tuning. We propose a practical, deployment-oriented two-stage workflow: rapid slide-embedding screening to prioritize robust representations and candidate endpoints, followed by patch-level training for high-value tasks where additional performance gains and interpretable regions are clinically worthwhile.
Building similarity graph...
Analyzing shared references across papers
Loading...
Dongheng Ma
Hinano Nishikubo
Tomoya Sano
Genes
Osaka City University
Building similarity graph...
Analyzing shared references across papers
Loading...
Ma et al. (Wed,) studied this question.
www.synapsesocial.com/papers/69d893406c1944d70ce04486 — DOI: https://doi.org/10.3390/genes17040371