Multi modal prediction of therapy response for rectal cancer patients

P02 will focus on the impact of the TME on survival of rectal cancer patients and develop machine learning methods that integrate molecular omics data to predict treatment outcome. The project is based on the notion that networks based on multi-modal data (histopathology and molecular omics) will outperform networks trained with only one of the two data modalities. Thus, P02 will couple multi-view data via graph neuronal networks where the individual nodes represent cells, which carry RNA expression and morphological features, and edges connect neighboring cells based on their Euclidean distances (cell-graphs).