A dark blue abstract futuristic digital landscape with light blue lines dots and stars on horizon.

Expo Talks: The future of AI-based diagnosis

When we think about Artificial Intelligence, our minds tend to wander to the likes of Google, Spotify and Meta – places with algorithms that we are all familiar with. Or else we’ll conjure up an image of a future filled with robots and automation. What doesn’t tend to immediately spring to mind is the way that AI is transforming healthcare, supporting diagnoses and informing the way that clinicians treat their patients.

In this episode of ‘Expo Talks’, Jamie Keenan, Associate Director of Health Unlimited chairs an eye-opening discussion on what Artificial Intelligence is and how it works in medical settings. He is joined by Professor Sotirios ‘Sotos’ Tsaftaris, who is not only Chair in Machine Learning and Computer Vision at The University of Edinburgh, but is the Canon Medical/Royal Academy of Engineering Research Chair in Healthcare AI. With them is Dr Ken Sutherland, President of Canon Medical Research Europe.

From the Edinburgh offices of Canon Medical Research Europe, Dr Sutherland’s team of computer scientists, software engineers and researchers work with some of the most eminent medical professionals in their fields to develop transformative deep learning imaging technologies. They explore some of the most devastating illnesses you can imagine, such as strokes and mesothelioma (sometimes called ‘asbestos cancer’, for which there is no known cure) to create AI-driven tools that may make early-stage interventions possible. With an ageing population and the rising cost of healthcare, is AI the future of rapid diagnosis?

The perfect storm for healthcare

“We’re a victim of our own success,” says Dr Sutherland. He refers to the fact that people are living longer, with additional care needs and health and social care requirements right through their seventies, eighties and nineties. This, coupled with the need to deliver treatment in the most cost-efficient way possible, as well as additional pressures brought by Covid 19, makes it the most challenging time ever for healthcare. “We’ve got this backlog of people who have unfortunately not been able to get care because facilities have been focused on pandemic response,” explains Dr Sutherland. “The situation two years ago was tough. Now, it's even tougher, and the pressure that individuals and the system are under in most healthcare delivery situations is incredibly challenging.”

How can AI help?

Dr Sutherland and his team have been working on training AI to recognise patterns of disease and have recently published findings on mesothelioma. “It creates this very unusual lesion around the lung,” explains Dr Sutherland. Working with clinical experts, he and his team have been investigating the impact of treatments on these lesions. “What we've created is an AI-based system that can measure that lesion, and therefore give the clinician some feedback on whether the treatment they’re attempting is actually having an effect positively to reduce that tumour or not.” Effectively, this provides clinicians with more, and faster, information on the conditions of their patients, as well as bringing additional data to the table for future diagnoses.

What are the challenges in bringing AI to diagnosis?

Professor Tsaftaris describes the way that machines learn and how each image they ‘see’ is essentially a sequence of numbers that will then be remembered, recognised and compared. “Over the last few years, less than a decade actually, we are seeing a rebirth of what we call ‘neural networks’,” explains Professor Tsaftaris. “These are computer programmes that learn patterns through examples.” However, the challenge lies in giving these programmes enough examples – particularly when the diseases that they are investigating are relatively rare and have changing characteristics. In this respect, Professor Tsaftaris believes that because of this a new kind of AI is required. “We will need to create a new AI that works in a different fashion. And there is now a new breed of AI that can work with less data. And it works by understanding what the data is about, rather than being provided with an example.”

“Turbo-charging clinicians”

How can humans and AI work together? There are clearly some aspects of patient care that are beyond the capabilities of machines, but Artificial Intelligence working in partnership with clinicians can take away some of the drudgery and empower them with more time to provide care and the ability to make swifter decisions. “What the computers will never replace is the actual decision-making, in my mind,” says Dr Sutherland. “The measurement of that lesion will be done automatically. But the interpretation of what that measurement means is still the key role for the clinician.”

Listen to the full discussion below and find out what the future is for AI in healthcare and learn more about the way that machines learn.

Through Canon Medical Systems Europe, we offer a full range of medical imaging solutions across the globe, including CT, X-Ray, Ultrasound and MRI. Canon Medical Research Europe generates breakthrough technologies and next-generation medical imaging software for these powerful diagnostic healthcare tools and leads the way in medical software research as development centre of excellence.

Written by Sarah Vloothuis, External Communications Senior Manager EMEA