Prof Clare Verrill
The Experimental Pathology subtheme focuses on translational cellular pathology research and is a novel collaboration with the engineering team based at the Big Data Institute. Cellular Pathology (Histopathology) is a discipline involving evaluation of tissue sections under a microscope and was highlighted within the UK Government’s Industrial Life Sciences Strategy (2017) as being ripe for innovation. Much of the workflow is manual and assessment of samples involves parameters which are unchanged for decades. This subtheme aims to bring novel technologies into the workflow, with a particular emphasis on computational pathology; bringing digital reporting into clinical practice and establishing libraries of slide images that are used to build novel artificial intelligence algorithms to improve workflow or enable correlation with molecular changes within tumours. The subtheme also provides core laboratory and biobanking support including: histology, immunohistochemistry, tissue microarray construction, whole slide imaging and image analysis.
Digital Pathology – Digitisation of Cellular Pathology Services and Deployment of AI
Oxford University Hospitals NHS Foundation Trust (OUH) has deployed a digital pathology platform in its diagnostic Cellular Pathology laboratory which has enabled full digitisation. We have moved from 0% digitisation in 2018 to full digitisation in 2020, which has been a fantastic multidisciplinary effort from the NHS histopathology team together with our Biomedical Science colleagues. We are one of the first histopathology laboratories in the UK to achieve this milestone. This was enabled in part by being part of the £14.4 million PathLAKE digital pathology consortium funded by Innovate UK and industry collaborators. Digitally enabled care is a core component of the NHS Long Term Plan.
There are potential efficiency savings and definite quality benefits to digital pathology, including quicker results to patients and increased availability of case sharing for expert opinion. Cases can also be shared across networks. We have also deployed slide scanners in some of our South 4 Pathology Partnership network, in Milton Keynes University Hospitals NHS Trust and Great Western Hospital Swindon. Digital review of these cases in Oxford expedites the patient pathway and review for multidisciplinary team meetings.
Digital pathology has been key to maintaining clinical histopathology services at OUH during the COVID-19 pandemic and enabled highly specialised services to continue in the face of social distancing and shielding. In addition to the clinical service, digital pathology has enabled us to continue training our junior colleagues via digital pathology, which they have found hugely beneficial.
Digital pathology creates a unique environment to test and deploy AI algorithms to support the work of pathologists, potentially leading to greater efficiency and safer services, for example, by AI double checking cases. We are looking to test and deploy algorithms that have been developed by our group and also commercially available regulatory cleared (for diagnostic use) algorithms. Although we are leading the way in testing such AI technologies, it is also early days for use in clinical practice and thus we are helping set national precedence around strategy and process to implement these technologies.
Digital pathology technology will be rolled to a further 10 sites (including the creation of a fully digital South 4 Pathology Network) across the UK during 2021-2023 with a further £13.5 million of confirmed grant funding from Innovate UK as part of the PathLAKE Plus programme.
Towards a common tissue-based marker of inflammation for Immune Mediated Inflammatory Diseases (IMIDs)
Strong evidence exists for the tissue leukocyte infiltrate as a pathological marker that can predict response to therapy such as checkpoint inhibition in cancer. By contrast, despite remarkable advances in complex technology that can measure markers of inflammation in tissue sections, there is often poor consensus on how to define the inflammatory burden of IMIDs at the level of the diseased organ, and generally poor validation across other local (imaging) or systemic measures of inflammation. Disease specific histopathological scoring systems are rarely transferable to other pathologies, making comparisons across diseases impossible. A new approach is therefore required in order to enable comparison of tissue-based inflammatory burden in the new generation of cross-disease therapeutic trials running through the A-TAP (Arthritis Therapy Acceleration Programme).
The key components of the inflammatory infiltrate are generic and measurable using simple histology, while its spatial organisation across tissue subcompartments may be both generic (in the case of ectopic lymphoid structure (ELS) formation) and tissue specific. AI-based machine learning systems are ideally placed to integrate the quantity and organisation of inflammatory infiltrates using standard diagnostic H&E sections and have been used to remarkable effect to classify tumour tissues into groups defining prognosis and therapeutic responsiveness.
The aim of the project is to produce an AI analysis approach to these cohorts in order to determine an index of tissue inflammation (ITI) that maps onto clinical or other parameters.