We apply deep learning to whole-slide images and pathomics to characterise brain tumours at the cellular level, turning microscopy into quantitative, reproducible diagnostic information.

Computational Digital Neuropathology brings the lab's quantitative approach to the microscope. Working with whole-slide images of brain tissue, we develop deep-learning and pathomics methods that detect, classify and characterise disease at the cellular and tissue level, turning what has traditionally been a qualitative, expert-driven read into reproducible, quantitative information. Our work includes automated analysis of tumour histology, extraction of pathomic features that capture cell morphology and tissue architecture, and models that link these features to diagnosis, grade and molecular status. A recurring theme is collaboration across institutions and modalities, bridging digital pathology with imaging and genomics so that a tumour can be understood across scales, from the whole brain down to the single cell. Because pathology sits at the heart of diagnosis, we place particular emphasis on robustness and generalisability: models that hold up across scanners, stains and sites, and that clinicians can trust.