Leveraging multi-modality data for targeted biopsy and risk stratification of patients with inflammatory Bowel disease

Inflammatory bowel disease (IBD) is characterised by a long-standing (chronic) digestive tract inflammation. Patients with IBD have a six times higher chance of colorectal cancer and require frequent monitoring. In the UK, half a million population are affected with IBD, with a 94% rise in incidence among adolescents and costing nearly 1.5 billion pounds per year to the NHS.

Studies suggest that by 2030, there will be a 60% increase in colorectal cancer. Leeds Teaching Hospital has approximately 5000 patients with IBD under examination, and about 5% of these patients need a colonoscopy to monitor cancer risk. Due to the complex organ and severity of the disease, surveillance is very subjective and often leads to missed cancerous lesions. Many biopsies (over 30) are required to detect cancerous changes, which is inconvenient to patients and costly.

I aim to develop novel artificial intelligence (AI) solutions that will – minimise subjectivity in clinical diagnosis, allow early identification of patients at higher risk of progression to cancer,and improve understanding of the disease and its progression.

My proposed AI methods will enable high-quality endoscopy in a reduced time. It will allow targeted biopsy, thereby reducing the number of biopsies without compromising patient outcomes. I will create a reference chart to help discuss cases among experts from different departments. Understanding IBD is essential but complex, as the inflammation of cancer has not been observed using data-driven approaches previously. This project will also link patient outcomes from multiple visits to train computers to understand disease progression.