Professor Adrienne Flanagan
University College London
Awarded: £50,000
Sarcoma UK’s funding is helping support a £1 million research project using the power of AI to help diagnose sarcomas.
The challenge
Early diagnosis is vital for patient care, as it guides treatment and informs prognosis. Connective tissue tumours such as sarcomas are rare with over 100 subtypes, making accurate diagnosis challenging. There is also a shortage of pathologists in the NHS and sarcoma classification is becoming increasingly complicated. This is leading to ever-increasing time and costs to deliver diagnosis, which can delay treatment.
How will this project tackle this challenge?
Building on a grant awarded by Sarcoma UK in 2022, the team will develop a bespoke artificial intelligence (AI)-based network to help diagnose sarcoma. An algorithm will prompt pathologists to request the relevant diagnostic test to confirm the suggested diagnosis. The algorithm can currently detect 15 soft tissue sarcomas, but the team aim to increase this to more than 50. AI-based tools can make diagnosis easier to verify, but they need to be trained with large numbers of images. The number of subtypes and their rarity makes this especially challenging in sarcoma.
First, the team will scan two decades’ worth of images from multiple NHS trusts using several different pathology slide scanners. The AI tool will then be tested by running prospective sarcoma cases through the model, which will then be used by AI developers to improve the model. Sarcoma UK’s funding will pay technicians to retrieve, scan and analyse slides from the archives.
What this means for people affected by sarcoma
An AI based model will help a diagnosis be reached more quickly, using fewer resources, expense, and staff time. If effective, it means that sarcoma patients will get a faster diagnosis, and therefore access the correct treatment more quickly. The image library will also provide a unique resource for future studies and training of new pathologists.
