Multicellular organisation requires the coordination of multiple signalling pathways that regulate cell shape as well as cell-cell and cell-microenvironment interactions. Such coordination is often lost in cancer resulting in changes in tissue architecture, uncontrolled growth, and metastasis. Imaging technologies provide a powerful approach to simultaneously study the signalling and organisational state of cells. I will discuss the development of machine learning and computer vision methodologies for automated identification of genetic programmes underlying tissue organisation. Importantly, these studies reveal that cell context and shape can modulate cell signalling even in isogenic cell cultures. Using orthogonal datasets and integrative approaches, we validate the clinical relevance of cell shape and context in patient prognosis.