Abstract PS3-04-15: Merlin: dissecting therapeutically relevant elements of the oncogenic microenvironment using multi-modal deep learning

Key Points

• The study aims to improve clinical decision-making and drug discovery by utilizing deep learning to analyze tumor microenvironment data.
• Curated a dataset of 3111 H&E whole slide images and clinico-genomic data.
• Developed attention-based multiple instance learning models for tumor microenvironment phenotyping.
• Validated findings with tissue-matched immunofluorescence and correlations with clinical outcomes.
• MERLIN effectively derived insights from H&E and clinical data, indicating tumor characteristics.
• Identified spatial patterns of lymphocyte clusters correlating with T-cell cytotoxicity.
• Established a link between T-cell activity and response to checkpoint inhibitors with a p-value of 0.017.