Image 2 (300 dpi credit needed)_edited_edited.jpg

Digitising Disease

Artificial intelligence under the microscope

Artificial intelligence (AI) is playing an increasingly important role in the way diseases are diagnosed. But what is meant by AI and what do patients need to know about its use in healthcare? This exhibition puts AI under the microscope, raising a series of questions posed by patients and the public concerning the use of AI when diagnosing disease.

 

What is Artificial Intelligence?

​COMPUTERS LEARNING COMPLEX TASKS

BY ANALYSING LARGE AMOUNTS OF DATA

 
Robot.jpg

Artificial intelligence, or AI, is predicted to change our world in many ways. From driverless cars and advanced robots to facial recognition systems and smart home devices, we are beginning to see that change all around us. But what is meant by AI?

 
board-453758_1280.jpg

There is no agreed upon definition of AI. When medical researchers use the term, however, they usually refer to a field of computer science and engineering, one concerned with programming computers to perform “intelligent” or complex tasks.

But some tasks are too complicated to program “by hand,” and in these cases computers can learn to write their own instructions for how to complete a task. “Machine learning” is the name given to this type of AI program.

 

“Deep learning” is a new AI process by which computers learn how to perform complex tasks.

 
Image Brain (credit needed).jpg

In deep learning, the computer is fed large amounts of data and looks for patterns by which it can make predictions regarding new data.


Deep learning techniques are modelled on how we understand the brain to work. The brain learns from repeated experience, which strengthens the connections between “neurons” or brain cells. With deep learning, the computer learns from data. And the more data it has, the better it gets at learning and performing the task.


The technology is already used in everyday life, for instance, when using online image searches. The question is whether that same technology can be used in medical research to recognise images of diseased tissue.

 

How could AI help diagnose disease?

PATHOLOGISTS WORKING WITH AI TO DIAGNOSE DISEASE

 
Image Bowel Cancer 3_edited.jpg

If a patient has symptoms of a disease, a “biopsy” (a sample of their tissue) may be taken. A specialist doctor, called a pathologist, then uses a microscope to view thin slices of the biopsy to see if they can identify patterns of disease. They also write a report explaining the diagnosis, how serious or advanced the disease is, and what treatments may work.

 
Digital 1_edited_edited.jpg

Some hospitals now use “digital” methods instead of microscopes, scanning the tissue slide so it can be viewed on a computer screen. These digital images can also be saved in a database, which can be used for training AI to find patterns in the data using deep learning techniques. Those deep learning technologies could, in the future, assist pathologists with various tasks, including:

- counting and measuring features in the tissue sample to assess how aggressive or advanced a disease is


- screening slides and drawing the pathologist’s attention to possible abnormalities.

 

What are the benefits of using AI for diagnosis?

 
 
More efficiency square

More efficiency

AI could improve the efficiency of diagnosis, allowing the NHS to examine slides more quickly and cheaply. This could lead to faster diagnoses and treatment. And by replacing some of the simple, repetitive tasks of diagnosis, AI would allow doctors to concentrate on the most urgent or complex cases.

What are the risks and risk protections?

 

Patients understandably have many questions and concerns about the adoption of new medical technologies in the NHS. 

 
pexels-steshka-willems-3018993 (1)_edited.jpg

"What happens if AI is commercialised?"

PUTTING PUBLIC HEALTH FIRST

Many people are concerned about commercial involvement in the NHS and how it could affect healthcare’s commitment to patient care. As with many tools developed for medicine, the NHS will likely need to involve commercial partners to develop, install, and support AI’s use in the hospital. Research and development of those tools, however, should be driven by public benefit, rather than profits.

Have  your say!

 

How AI should be brought into medicine is a question that concerns not just doctors and computer scientists, but all of us. As a consequence, it is up to all of us to decide how it should be used. The best way to shape the future medical use of AI is to be part of the discussion. We have 3 ways you can be involved:

You can complete our survey to let us know what you think about the use of AI for diagnosis.

You can apply to take part in our upcoming online citizen’s panel, during which members of the public will debate the pros and cons of AI in diagnoses to determine its future development

If you have specific questions that come up from reading the exhibit, please contact Bethany Williams at NPIC@leeds.ac.uk or visit us at www.npic.ac.uk for more information.

 

Acknowledgements:

This exhibition was generously supported by National Pathology Imaging Co-operative and the Wellcome Centre for Ethics and Humanities at the University of Oxford.
 

The National Pathology Imagining Cooperative (NPIC) is a consortium of academic, industry and NHS partners who are digitalizing pathology services across the UK in order to build a system for driving artificial intelligence research in the future. 
The ‘National Pathology Imaging Co-operative, NPIC (Project no. 104687) is supported by a £50m investment from the Data to Early Diagnosis and Precision Medicine challenge, managed and delivered by UK Research and Innovation (UKRI)

The Wellcome Centre for Ethics and Humanities is supported by funding from the Wellcome Trust (Grant no 203132).


Contributors:

The exhibition was co-curated by: Francis McKay, Nina Hallowell, Milly Farrell, Nick Pitt, Bethany Williams, Graham Prestwich, Eloise Pearson, Alex Wright, Darren Treanor, Derek Magee, and the members of NPIC’s Patient and Public Advisory Group. Additional advice was provided by Kelly Richards (Oxford University Museum of Natural History) and Michael Fulton (Leeds Teaching Hospital Trust). Panels were designed by Francis McKay and Gemma Hattersley.