September 10, 2025Nucleic Acids ResearchOpen Access

FmH2ST: foundation model-based spatial transcriptomics generation from histological images

View Full Paper

Ask AI

Bookmark

Authors

Discussion

Foundation model-based method predicts gene expression patterns in tissues, highlighting spatial heterogeneity and complexity.

FmH2ST significantly improves spatial transcriptomics (ST) predictions from histological images, enhancing gene expression modeling.
By utilizing a dual-branch framework, FmH2ST effectively captures inter-slice heterogeneity and intra-slice complexity, outperforming existing methods.
Its multilevel feature extraction and graph neural network techniques enhance the quality of gene expression predictions.
FmH2ST's predictive capabilities could support advancements in clinical diagnostics and tailored personalized treatments.

Wang et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark