Deep learning predicts gene expression from tissue images

Researchers have developed and validated deep regression models capable of predicting gene expression directly from whole-slide images (WSIs) of cancer tissue. These models, utilizing attention-based multiple instance learning and pathology foundation models, demonstrated strong performance across multiple datasets. External validation on independent cohorts confirmed the models' ability to generalize and recover clinically relevant molecular information, such as PAM50 gene sets, retaining prognostic value for patient survival. AI

IMPACT This research demonstrates a scalable approach for transcriptomic phenotyping and risk stratification, potentially reducing costs and increasing accessibility in precision oncology.

RANK_REASON Academic paper detailing a new methodology and its validation. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• Fredrik K. Gustafsson
• TCGA

• paper
• model release
• other

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