Current students

Inflammatory Bowel Disease (IBD) refers to a group of chronic inflammatory conditions affecting the gastrointestinal tract, profoundly impacting patients' quality of life. This complex disease, typically managed with drugs that mitigate symptoms and, in some cases, surgical interventions, requires continuous monitoring. Despite extensive research, the pathogenesis of IBD remains only partially understood. However, it is known to primarily encompass two disorders: Ulcerative Colitis (UC) and Crohn's Disease (CD), each with unique clinical presentations but unclear molecular and genomic features.Our project aims to use an integrative approach that combines Whole Slide Images (WSI), genomic data, and spatial transcriptomics. WSIs offer a comprehensive view of tissue architecture but pose computational challenges due to their large size and complex structure. Addressing this, we are developing a pipeline to extract cell types and tissue substructures from WSIs, enabling us to understand the hierarchy of tissue organisation and its relationship to disease.A human-in-the-loop approach will be exploited where pathologist annotations are used to refine segmentation model predictions. This in turn will enable us to obtain imaging-derived phenotypes (IDPs) which can be used to perform Genome Wide Association Studies (GWAS) linking changes in specific histological morphology to variants present in the genome.Moreover, we will use the obtained segmentation masks with spatial transcriptomics to delineate specific transcriptional profiles of UC, CD, and non-affected controls. Through this, we aim to identify distinct transcriptional signatures that define each condition and their specific histopathological entities (e.g. crypt abscesses, architectural distortions, granuloma foci, etc.).Ultimately, we will employ advanced machine learning techniques to create a unified, low-dimensional representation of these morphological features and their transcriptional signatures. This approach will illuminate the disease-specific changes and advance our understanding of the genotype-phenotype landscape in IBD. By integrating these different modalities, we want to significantly enhance the accuracy of IBD prognostication, paving the way for personalized treatment strategies and improved patient outcomes.