New AI model predicts gene expression from histology images

Researchers have developed HierarchicalDAEW, a novel dual-graph architecture designed to predict gene expression from H&E histology images. This method addresses the limitations of current spatial transcriptomics assays by integrating tissue architecture and quantifying prediction reliability. The system utilizes a domain-aware edge-weighted convolutional operator and a gene-level graph that combines protein-protein interaction priors with tissue-specific co-expression, achieving superior correlation with ground-truth expression across various human tissue types. AI

IMPACT This model could enable more accessible and routine transcriptome-wide profiling by leveraging standard histology images.

RANK_REASON Research paper detailing a new AI model and methodology. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• arXiv
• H&E histology
• HierarchicalDAEW
• Kritanu Chattopadhyay
• Leiden clustering
• spatial transcriptomics

• paper
• model release
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →