The recent Life Science Sector Deal placed a strong emphasis on digital pathology… but does AI really have a role here, and will the government’s hope for a successful partnership with industry players work? We hear from two key players in the new Sector deal landscape with some answers…
Professor Manuel Salto-Tellez
Chair of Molecular Pathology at the Precision Medicine Centre of Excellence, Queen’s University Belfast. He is also a scientific advisor to Philips Computational Pathology.
Since 1953, when Watson and Crick first described the double-helix structure of DNA, scientists have been applying knowledge of the molecular basis of disease to technological advances in molecular interrogation of clinical samples for the diagnosis and therapy of diseases.
A result is molecular medicine, arguably the fastest growing area in modern medicine. Accordingly, concepts such as molecular diagnostics, molecular-targeted therapies and therapeutic pathology are infiltrating daily practice.
The molecular medicine revolution, together with the ease with which we now generate patient data at different levels, has brought a key paradox: the generation of huge multimodal data sets with reasonable robustness is not difficult, but how do we manage and analyse this information? With information coming from many sources and levels (from genomics, pathology imaging, radiology-related imaging and clinical records to health information on social media) the management question is driving a medicine revolution and may offer crucial support mechanisms for evidence-based medicine.
The recent Life Sciences Industrial Strategy report published by the Office for the Strategic Coordination of Health Research is at the heart of this analysis. The document asks the UK to “host centres of excellence that provide support for specific medtech themes, focussing on research capability in a single medtech domain such as orthopaedics, cardiac, digital health or molecular diagnostics”. The document had an immediate effect: in September of 2017, NHS England chief executive Simon Stevens announced that the NHS will set-up two-to-five regional Digital Innovation Hubs, each covering regions of 3-5 million people. Finally, the race for the exploitation of anonymised clinical data has started.
More than meets the eye…
Of all the areas of medical artificial intelligence or “big data”, digital pathology probably provides the biggest challenges and growth opportunities. The core of traditional pathology has not changed since the days of Rudolph Virchow (1821-1902), i.e. scraping the surface of the information that different cell types, provide by virtue of their 2-dimensional or 3-dimensional arrangements.
However, tumour microenvironments, cancer ecology and morpho-molecular heterogeneity are concepts that are beginning to explain the perceived degree of information held by tissues and cells still untapped by the human eye.
Our vibrant academic environment in the UK is the best soil for a new healthcare revolution. UK academia has already created extraordinary products at the heart of this new development. QuPath, an open source digital pathology analytical system is one good example. Developed at Queen’s University Belfast (QUB), QuPath has had more than 5,000 downloads worldwide in under a year, and is becoming a reference tool for digital pathology analysis.
However, open source solutions do not enrich the fabric of biotech industry in isolation. More than ever, the academia-healthcare-industry axis becomes extremely relevant to move this area forward. We believe that this relation is not linear, but a circular economy. This is particularly true for digital pathology, which both serves very specific needs for routine pathology and, at the same time, brings levels of information that the human eye has never been able to perceive. So even if the pathologist doesn’t generate this information (algorithms do), they can integrate that information in the overall diagnostic and therapeutic decision-making.
A story that epitomises how these changes may actually be realised in the future for digital pathology started a few years ago in QUB. The university has a distinguished track-record for incubating companies. One of them – PathXL – became a globally leading UK-based SME in the digital pathology sector, led by visionary individuals such as Peter Hamilton and Des Speed. The Northern Ireland Molecular Pathology Laboratory (today the QUB’s Precision Medicine Centre of Excellence), with scientists at PathXL, developed a key AI algorithm to automatically recognise and characterise areas of cancer in histological sections ahead of molecular testing. The algorithm became one of the leading products (“TissueMark”) of the company and now uses the most advanced deep learning algorithms to detect, annotate and analyse cancer in pathology samples. PathXL was subsequently acquired by Philips Digital Pathology Solutions.
This product is now back in our laboratory as a tool for sample selection in diagnostic molecular testing, offering a model of the role industry should have in these early days of digital pathology application into healthcare, and a reminder of the cyclical nature of healthcare-academia-industry interactions.
It has been said that our NHS represents the best milieu for healthcare-industry interactions. It is fair to say this can be supported by excellent examples while, simultaneously, being a far from global truth across the NHS. Perhaps digital pathology offers another opportunity to show how the NHS could be a global engine for innovation, a development that would allow people to live longer, better, lives.
[box type="shadow" align="aligncenter" ]Life Sciences Sector Deal – Government and industry commitments
Government commitment includes: raising the intensity of research and development in the UK; strengthening the environment for clinical trials; supporting the growth of medicines manufacture; improving the UK environment for businesses with the potential to scale up; implementing the Accelerated Access Review and supporting improvements to the UK’s health data infrastructure.
Industry commitments include: investing in the UK; closer collaboration between companies and academia; collaborating with the Government on advanced health research projects and match-funding for Industrial Strategy Challenge Fund Investment. [/box]
Director of Healthcare Development and Strategic Services at Roche Diagnostics in the UK and Ireland
As part of its strategy to ensure a thriving economy post-BREXIT, the Government is keen to capitalise on it’s unique position: our universities and research institutes rank among the best in the world; the UK is home to some of the most successful global life science companies and of course our NHS. To realise the full potential, the Government recognises that it needs to work with universities, charities and business as set out in the Life Sciences Industrial Strategy. The Life Sciences Sector Deal announced in December 2017 was another positive step forward and we at Roche Diagnostics are proud to be working in partnership with all interested parties to help maximise the opportunities. In fact, we see partnership and collaboration as not only desirable but necessary in developing a sustainable NHS.
The Life Sciences Sector Deal sets out a series of commitments, both on the side of the Government and industry. This is all underpinned by a plethora of partnerships and collaborations and whilst in their early stages, the continuing strengthening and building of partnerships will be critical to seeing all of these commitments becoming reality. And progress will be scrutinised by an independent Industrial Strategy Council as yet to be appointed.
The Life Sciences Industrial Strategy published in August, recognised the need to consider and fund projects that will impact on the direction of healthcare delivery over the next twenty years. It proposed that some should be large-scale infrastructure projects underpinned by novel technology and higher-risk science – the intention being to create commercial success by leading and developing new industrial sectors. These broad classes of scientific opportunities have been labelled as the Health Advanced Research Programmes (HARP).
As part of this, we are working to digitise pathology glass slide images, creating data repositories that enable the opportunity to create artificial intelligence based algorithms. This transformational change to the way in which images are viewed, stored and analysed will help deliver faster and more accurate results to patients. Especially in the case of colorectal or lung cancer where patients all too often present late, turnaround time for test results is critical. Digitisation will mean that experts from any location will be able to view the slide images without delay. And with the current shortages of pathologists, they will be able to better use their time reviewing the more complex cases.
Additionally, making more information available electronically opens possibilities for the discovery of new treatments and the development of artificial intelligence algorithms in pathology diagnosis, paving the way for more effective and better outcomes for patients, as well as spawning a new industry in the area of machine learning.