AI-predicted spatial transcriptomics unlocks breast cancer biomarkers from pathology.

AI model predicts spatial gene expression from histopathology slides in breast cancer, identifying subgroups with distinct survival outcomes and improving treatment predictions.

• Why it matters: Understanding tumor heterogeneity and microenvironment is crucial for personalized cancer therapy, but current spatial transcriptomics methods are costly and limited in scale.
• What they did: The study developed "Path2Space," a deep-learning approach trained on extensive breast cancer data, to accurately predict spatial gene expression and cell-type distributions from standard pathology images.
• The result: This method enables cost-effective, large-scale biomarker discovery and improves prediction of patient responses to therapies, facilitating translational research and potential clinical applications across cancers.