Improving health with artificial intelligence
Developing AI algorithms for early diagnosis and personalised treatment of cancer and other diseases
The Tissue Image Analytics Centre at the University of Warwick is renowned for its leading research in computational pathology. Through a consortium of universities, hospitals and industry, the research team at Warwick led by Professor Nasir Rajpoot has helped establish a national centre of excellence for artificial intelligence (AI) in pathology, focused on housing a large-scale data lake – a centralised and accessible repository of clinical pathology imaging data for AI researchers. The societal benefits from this initiative, called PathLAKE (Pathology Image Data Lake for Analytics and Education) are numerous, including transforming NHS pathology by improving diagnostic efficiency and personalised medicine.
Pathologists currently undertake testing via visual examination of pathology specimens under the microscope, a process that is inherently subjective. A large proportion of tissue samples examined are normal and using specialist pathologist time to establish this is expensive. In addition, pathologists' performance can be variable and is limited to human expertise. This is particularly noticeable where complex tasks requiring analysis of thousands of minute data points are needed to decide the optimal form of treatment. Even in the very best centres this can mean patients with cancer may receive expensive and unpleasant treatment they will not benefit from or fail to receive treatment which they need. AI tools not only save valuable pathologist time that can be used elsewhere but can also aid pathologists’ decision-making.
The ongoing research led by Professor Nasir Rajpoot and colleagues focuses on two areas:
- Diagnostic efficiency - AI solutions that can introduce automation and improve diagnostic efficiency, reduce pathology workload, lower cost to NHS and help address the well-known pathology workforce issues (in the UK and globally).
- Discovery - prognostic/predictive markers for personalised medicine - AI tools for tissue-based markers for immuno-oncology, predictive markers of chemotherapy response in breast cancer and in prostate cancer, development of novel histology-based subtypes and markers for progression in active surveillance.
Through the digitisation of five major NHS laboratories and the creation of a data lake, the PathLAKE consortium has enabled rapid validation and deployment of AI tools which are expected to improve workflow efficiency (>15%) and clinical care pathways, covering 46% of the population of England and Wales. Early diagnosis and early treatment of cancer is known to lead to improved outcomes, by improving objective predictors for cancer relapse based on automated tissue analytics, it will be possible to provide more adequate treatment at an earlier stage, thereby reducing mortality rates. Additionally, PathLAKE is expected to drive growth in the UK economy by generating joint intellectual property and promote innovation by UK SMEs. The development of next generation AI algorithms will establish a powerful ecosystem of UK partnerships as well as nurturing new international partnerships. PathLAKE also provides a network for multi-site clinical trials not only for advancing research projects but providing world-class training of pathologists in digital pathology.