Supporting Medical Research

We think that machine learning technology, a type of artificial intelligence, can bring huge benefits to medical research. By using this technology to analyse medical data, we want to find ways to improve how illnesses are diagnosed and treated. Our goal is to help clinicians to give faster, better treatment to their patients and all our research work is done in collaboration with doctors and professional healthcare researchers.

It’s very early days in our health research programme but we want to share information about our current medical research work. All research projects go through rigorous regulatory and Trust approvals and are conducted only on non-identifiable patient data. You can read more about the healthcare research permissions process on our Information Governance page.

Eye Diseases

Two million people are living with sight loss in the UK, of whom around 360,000 are registered as blind or partially sighted.

At the moment, eye health professionals rely on digital scans of the eye to diagnose and determine the correct treatment for common eye conditions such as age-related macular degeneration and diabetic retinopathy.


This highly innovative and exciting research is so crucial... it is a future illuminated with hope.”

Elaine Manna, patient at Moorfields
fullscreen fullscreen_mobile
An OCT scan of one of the DeepMind Health team's eye

These scans are highly complex and to date, traditional analysis tools have been unable to explore them fully. It also takes eye health professionals a long time to analyse eye scans, which can have an impact on how quickly they can meet patients to discuss diagnosis and treatment.

Our research project is investigating how technology could help to better analyse these scans, giving doctors a better understanding of eye disease. We hope this will lead to earlier detection and treatment for patients and ultimately help to avoid cases of preventable eye disease.

play DeepMind Health – Moorfields Eye Hospital London Collaboration

Professor Sir Peng Tee Khaw, Director of the National Institute for Health Research Specialist Biomedical Research Centre in Ophthalmology at Moorfields Eye Hospital NHS Foundation Trust and UCL Institute of Ophthalmology"Our research with DeepMind has the potential to revolutionise the way professionals carry out eye tests and could lead to earlier detection and treatment of common eye diseases such as age-related macular degeneration. With sight loss predicted to double by the year 2050 it is vital we explore the use of cutting-edge technology to prevent eye disease."

Dr Dolores Conroy, Director of Research at Fight for Sight, the UK’s main eye research charity welcomes this partnership and comments "We are really excited about this collaboration and the potential of machine learning to analyse the thousands of retinal scans taken each week in the NHS allowing eye health professionals to make faster, more accurate diagnoses and more timely treatments thus preventing sight loss. In the longer term this technology could provide important insights into disease mechanisms in wet AMD and diabetic retinopathy."

Cathy Yelf, Chief Executive of the Macular Society: "This is an exciting development towards early detection of eye disease and finding a cure for conditions including age-related macular degeneration (AMD). AMD is a devastating condition and delays due to pressure on eye clinics have resulted in some people suffering unnecessary sight loss. This technology could ease that pressure if it can accurately diagnose conditions such as wet AMD resulting in urgent referrals for only those that need them."

Clara Eaglen, RNIB Eye Health Campaigns Manager: "AI technology that can check retinal scans and detect eye disease at a much earlier stage could play a big role in tackling avoidable sight loss. In many cases, once sight is lost it cannot be restored, so earlier detection that leads to rapid treatment will be hugely beneficial. We look forward to seeing the results of the work as the research progresses."

Radiotherapy Planning

1 in 75 men and 1 in 150 women will be diagnosed with oral cancer in their lifetime, and head and neck cancers affect over 11,000 people in the UK every year. Yet these and other cancers in the head and neck are particularly difficult to treat with radiotherapy.

The high number of delicate structures and nerves concentrated in this area of the body, as well as their proximity to the brain, means that radiotherapy must be painstakingly planned to ensure no healthy structures are damaged, in a process called segmentation. This involves producing a detailed map of the area to be treated, isolating cancerous tissue to be treated from healthy tissue.

Our research project investigates whether machine learning could speed up the segmentation process while maintaining its accuracy. Currently, segmentation can take up to four hours for head and neck cancers. In so doing, we hope to show how clinicians’ time could be freed up to focus on patient care, education, and research as well as contribute to a radiotherapy segmentation algorithm that could potentially be applied to other areas of the body.

fullscreen fullscreen_mobile

Dr Yen-Ching Chang, clinical lead for radiotherapy at UCLH, said: "This is very exciting research which could revolutionise the way in which we plan radiotherapy treatment.

“Developing machine learning which can automatically differentiate between cancerous and healthy tissue on radiotherapy scans will assist clinicians in planning radiotherapy treatment. This has the potential to free up clinicians to spend even more time on patient care, education and research, all of which would be to the benefit of our patients and the populations we serve.

“This collaboration also means our patients continue to benefit from the most cutting-edge developments in healthcare technology.”

Professor Kathy Pritchard-Jones, chief medical officer of London Cancer, the integrated cancer system that serves a population of more than 3.5 million, said: "Head and neck cancer is rare and is one of the most complex tumour sites to treat. Therefore, if we can develop technology to assist in planning radiotherapy treatment for these tumours, we would expect that such a breakthrough would be transferrable to other types of cancer. This would not only benefit UCLH patients, but patients across the country.”

Dr. Anna Thompson, Consultant Oncologist at UCLH, said: “The NHS is always looking at ways of improving patient pathways and experience. As an UCLH Oncology Consultant who plans radiotherapy for the treatment of patients with cancer, I am very excited to work with DeepMind exploring new technologies that could speed up the radiotherapy planning process. This could free up doctors and radiographers to spend more time with patients and further develop the service.”

Ruheena Mendes, Consultant Oncologist at UCLH, said: "This is an exciting, bold project that will aid clinicians to provide  excellence in radiotherapy planning and treatment.

"The success of the project will provide patients with highly accurate treatment with a quick turnaround between diagnosis and commencement of treatment and has the potential to impact positively on outcomes.

"With an increase in the incidence of virally driven head and neck cancer and the advanced technology of targeted radiation delivery, there is call for accurately defining tumour and surrounding organs in a streamlined fashion in radiotherapy."

Who we're working with

Moorfields Eye Hospital NHS Foundation Trust

Moorfields Eye Hospital NHS Foundation Trust is one of the leading providers of eye health services in the UK and a world-class centre of excellence for ophthalmic research and education.

The Details

Digital scans of the eye are non-invasive imaging tests that help assess the retina, the tissue lining the back of the eye. These scans allow eye health professionals to help diagnose and provide treatment guidance for serious eye conditions including age-related macular degeneration and diabetic retinopathy.

  • Diabetic Retinopathy
    Up to 50% of people with proliferative diabetic retinopathy who do not receive timely treatment will be registered as blind within five years. Up to 98% of severe vision loss from diabetic retinopathy can be prevented by early detection and treatment. By efficiently analysing the large number of scans and images taken of the eye every year, a machine learning algorithm could ensure the right patients in need of treatment are seen at the right time by the right clinician.
  • Age-related macular degeneration
    Age-related macular degeneration (AMD) will affect an estimated 2.5 million adults by 2020 in the UK alone. Around 15% of early cases of AMD progress to the more serious form of wet AMD, which can be difficult for clinicians to predict. Machine learning has an opportunity to filter through the huge quantity of information available from scans of the eye, and potentially recognise subtle features which predict the onset of wet AMD. This early warning system could initiate faster intervention and the prevention of patient eye deterioration in the future.

We’re working with Moorfields to apply machine learning algorithms to automatically detect and segment eye scans. Moorfields has supplied DeepMind with permission to access to their database of anonymised digital images of the eye, and to use that data to perform the research.

References

University College London Hospital NHS Trust

University College London Hospital has one of the largest centres for head and neck cancers in England and is a world leader in oncology research.

Head and neck cancers

Head and neck cancers affect over 11,000 people a year in the UK. Radiotherapy is used as first-line treatment in 40% of cancers, but this rate is higher for cancers of the head and neck, for which most patients will receive radiotherapy in their treatment.

Cutting-edge machine learning algorithms allow the creation of sophisticated image recognition tools. Using those tools on CT and MRI scans from UCLH patients suffering from a cancer of the head and neck, we hope to improve the efficiency of the complex treatment planning that is required to ensure enough radiation is given to the tumour.

UCLH has supplied DeepMind with permission to access a set of up to 700 scans from former patients who have consented to their data being used for medical research.

Moorfields Eye Hospital London

  • How did the partnership with Moorfields come about?

    Pearse Keane, a consultant ophthalmologist at Moorfields Eye Hospital, approached DeepMind via our website to explore how our technologies could help to analyse scans to provide a better understanding of eye disease. If you have an idea for medical research you’d like us to collaborate with you on, please get in touch via sayhi@deepmindhealth.com.

  • What will the project involve?
    Eye health professionals use scans of patients’ eyes to detect and diagnose serious conditions and diseases. Many thousands of eye scans are performed around the UK every day, both in hospital eye clinics and in the community. For example, more than 3,000 optical coherence tomography (OCT) scans are performed every week in Moorfields Eye Hospital alone.

    OCT scans are highly complex and require specialised training for doctors and other eye health professionals to analyse. As a result, there are often significant delays in how quickly patients can be seen to discuss their diagnosis and treatment. To date, traditional computer analysis tools have been unable to solve this problem

    Our research project aims to investigate how machine learning technology could help to analyse these eye scans (OCT and fundus images), giving eye care professionals a better and faster understanding of eye disease.

  • What is the project aiming to achieve?

    Ultimately, we want to help eye health professionals to make faster and more accurate diagnoses of eye conditions and diseases, so they can diagnose and treat patients as early as possible.

  • What’s the difference between OCT and fundus images?

    Traditional retinal images, also known as fundus images, are basically photographs of the back of the eye (the retina). OCTs are a cross-section of the retina, giving eye health professionals a more detailed picture of any damage, which can allow for earlier diagnosis of conditions such as diabetic retinopathy and age-related macular degeneration.

  • Where can I find more detail on the agreement between DeepMind and Moorfields?

    You can request a copy of the research collaboration agreement (with minor redactions for commercial sensitivity) and the ROAD (research on anonymised data) form which was approved by Moorfields Research and Development department by emailing press.office@moorfields.nhs.uk.

    You can also view a copy of a research protocol we've written for this project on the open access website F1000Research.

  • How long will the project last?

    The research project agreement is for five years, though either party can terminate early with 30 days’ notice. At the end of the agreement, DeepMind must destroy all copies of data received through the agreement.

  • How much data has DeepMind been given access to?

    Over the course of the research project, Moorfields Eye Hospital will share approximately one million anonymised digital eye scans, used by eye health professionals to detect and diagnose eye conditions. Anonymous clinical diagnoses, information on the treatment of eye diseases, the model of the machine used to acquire the images and demographic information on age (shown to be associated with eye disease) will also be shared during the course of the project.

Pearse Keane, a consultant ophthalmologist at Moorfields Eye Hospital, approached DeepMind via our website to explore how our technologies could help to analyse scans to provide a better understanding of eye disease. If you have an idea for medical research you’d like us to collaborate with you on, please get in touch via sayhi@deepmindhealth.com.

  • What approvals has DeepMind been given for this research project?

    DeepMind has been given permission for data access via a Research Collaboration Agreement with Moorfields Eye Hospital, and an approval to carry out research from the Moorfields Research & Development team through their Research On Anonymised Data (ROAD) approval pathway.

  • Do patients have to give their consent for their data to be used?

    The data used in this research is not personally identifiable. When research is working with such data, which is anonymous with no way for researchers to identify individual patients, explicit consent from patients for their data to be used in this way is not required. (For more information please refer to the ICO code of conduct.)
    As with all research collaboration agreements with non-NHS organisations, patients can opt out of any data-sharing system by contacting the trust’s data protection officer. Full details are available on the Moorfields Eye Hospital trust website. Patients will need their NHS or medical records number. Opting out of research applies to all research projects at Moorfields, not just the DeepMind collaboration.

    In this case, all future research will take place excluding those who have opted out of the study. As with all studies on anonymous datasets technical, measures will be adopted to render it impossible to identify which patient has opted out. It is therefore not possible to opt-out of research already underway for this reason.

  • Will any further patient information be shared between Moorfields and DeepMind in the future?

    DeepMind Health and Moorfields Eye Hospital are currently seeking approvals to allow the research to connect anonymous scans over time. This will allow further investigation of how eye diseases progress and the effects of treatment on eye conditions. As with research that is currently underway, all patient identifiable data will be removed before transfer.

  • How can patients be sure that no identifiable data is being shared with DeepMind?

    Anonymisation procedures are thoroughly validated and formally approved by the Moorfields Eye Hospital information governance team before any data transfer to DeepMind takes place.

  • Who owns the image database?

    Ownership of the database remains with Moorfields Eye Hospital throughout the agreement and afterwards.

  • Who owns the software that is being used and developed in this project?

    Artificial intelligence software developed by DeepMind during the research project is owned by DeepMind and a licence is granted to use the data provided, only to the extent it is incorporated into any software developed from the research. This licence continues after termination, however access to the database itself does not.

  • What processes are in place to ensure the data transferred to DeepMind is only ever seen by the research team?

    A data custodian has been appointed by DeepMind to control access to the data. Only those who require access to conduct the research work will be granted access. All researchers who are involved in the study are required to complete Health and Social Care Information Centre (HSCIC) and internal DeepMind information governance training before beginning research work.

DeepMind has been given permission for data access via a Research Collaboration Agreement with Moorfields Eye Hospital, and an approval to carry out research from the Moorfields Research & Development team through their Research On Anonymised Data (ROAD) approval pathway.

University College London Hospital NHS Trust

  • What does UCLH’s research agreement with DeepMind Health involve?

    Under the agreement, UCLH will provide DeepMind Health secure access to anonymised CT and MRI scans of approximately 700 head and neck cancer patients. All patients have consented to their data being used for research purposes.

  • What is the purpose of the research?
    The purpose of the research is to develop technology which can automatically identify and differentiate between cancerous and healthy tissues on CT and MRI scans of head and neck cancer patients to help target radiotherapy treatment.

    At present, this process, known as segmentation, can take clinicians up to four hours to complete manually, as tumours in head and neck patients are situated in extremely close proximity to healthy structures such as the eyes and nerves.

    The research aims to develop artificial intelligence technology to assist clinicians in the segmentation process so that it can be done more rapidly but just as accurately. Clinicians will remain responsible for deciding radiotherapy treatment plans but it is hoped that the segmentation process could be reduced from up to four hours to an hour.

    Longer term this has the potential to free up clinicians to spend even more time on patient care, education and research, all of which would be to the benefit of UCLH patients and the populations they serve.

    In addition, given that head and neck cancer is one of the most complex tumour sites to treat, if we can develop technology to assist in planning radiotherapy treatment for these tumours, we would expect that such a breakthrough will be transferrable to the other types of cancer. This would not only benefit patients at UCLH, but patients across the country.

  • If the research is successful, will the technology being developed replace the doctor’s role in deciding treatment?

    No. We hope the technology will assist clinicians but they will remain responsible for deciding patients’ treatment plans.

  • How long is the research project with DeepMind Health?

    The research project will last five years. Either UCLH or DeepMind Health can end the project early with 30 days’ notice.

  • What will you do with the results of the research?

    DeepMind Health will publish any results through normal academic channels, subject to formal peer review process.

  • Do the images being used for research contain patient identifiable information?

    No, all of the images will be anonymised by UCLH before being transferred to DeepMind Health for the research to begin.

  • Is any other information being used for this project?

    In addition to the radiotherapy planning scans, UCLH is also providing additional data attached to the anonymous scans relating to approximate age (in 5-year age groups), anatomy location, cancer type and radiotherapy received. This provides important context for algorithm development in helping to understand what the images show and how technology may be used to improve the planning process. Like the scans, this information is anonymised.

  • Does the patient data used in the project relate to former or current patients?

    The research involves anonymised scans dating back to 2008 of head and neck cancer patients who have since completed radiotherapy treatment. At the time, these patients would have consented to their anonymised data being used for research purposes. Scans of patients currently undergoing radiotherapy treatment will not be included in the research.

Under the agreement, UCLH will provide DeepMind Health secure access to anonymised CT and MRI scans of approximately 700 head and neck cancer patients. All patients have consented to their data being used for research purposes.

The data used in this research is not personally identifiable, it is anonymised. In these circumstances, consent from patients for their data to be used in research is not required and is covered by UCLH’s privacy statement.

When radiotherapy patients begin treatment, however, UCLH do ask them to sign a consent form which allows their anonymised data to be used for research purposes. Only patients who have given their consent for their anonymised data to be used for research purposes will be included in this study.