Maximizing the Potential of AI in Clinical Medicine

In this article, Dr Ken Sutherland, gives his perspective on what meaningful innovation should entail when it comes to AI. He discusses where we currently stand with AI and touches on critical points of attention needed to move deeplearning algorithms out of the research environment into real-world clinical practice.

How has artificial intelligence evolved so far?

Years of technological innovation has meant that ‘off-the-shelf’ computing equipment has become so powerful that AI deep learning (the process of generalizing patterns from large numbers of datasets) is possible at a reasonable cost. The availability of good quality, well-curated datasets has also improved, and new methods of AI learning have been invented that are better at mimicking humans. Combining these advances has enabled the application of AI to the entire Radiology workflow, and this change has the potential to be transformational. We can all learn a lot from the use of AI in the research environment.

From research to practice: What are the caveats?

Caution is required. A great deal is now understood about the pitfalls of training AI with poor-quality or biased datasets and the merits of using a truly representative population for verification.

In truth, the term ‘Artificial Intelligence’ as we use it in Radiology and Healthcare today is a misnomer. The AI algorithms being developed don’t understand the data that they use or the results that they produce. Algorithms alone have no understanding of pathology, disease, patients, or even care, but they are definitely useful given the challenges facing healthcare providers. Particularly during a time when we are emerging from a global pandemic that has changed medicine forever.

Healthcare professionals are becoming increasingly ‘AI savvy’ and are asking the right questions to industry and partners. AI algorithms that can support humans but not replace them is an achievable and desirable goal for all parties. Increasingly AI researchers are being challenged to demonstrate that their innovation works within a real workflow and not just in the test environment in the laboratory.

Towards the future: How can we optimally deploy AI in healthcare?

The appropriate use of AI to streamline the entire radiology workflow, from patient positioning on the scanner through to the final diagnostic process, can free-up professionals to spend more time with patients and create more time for non-routine work that demands their experience and skill.

Embedding AI into the clinical environment to create data driven workflows, in which the relevant information is provided to the clinician at the right stage of the process to enable optimal decision-making, is probably the final challenge that requires vendors to work together. No individual organization or company can currently deliver these next generation ‘smart workflows’ alone, but the goal of delivering improved healthcare for all is so compelling that alliances and ecosystems are forming to tackle this ultimate challenge at such a critical time.
We can harness AI now for the benefit of us all. It is already creating a positive difference in in image quality, reconstruction speed, enabling reduced radiation dose and accelerating workflows. However, the human expertise of clinicians remains essential and always will.

“Deployed effectively, AI can free-up healthcare professionals to spend more time with patients and create more time for non-routine work that demands their experience and skill.”

Dr. Ken Sutherland, President of Canon Medical Research Europe.
Dr. Ken Sutherland
Dr Ken Sutherland is President of Canon Medical Research Europe and is responsible for all aspects of the strategic development and operational control within Canon Medical Research Europe Ltd and is also a key member of Canon Medical Systems Global R&D leadership team. He serves as a member of the Scottish Government Inward Investment Forum, where he previously a lay member of the court of the University of Glasgow, UK, and as an advisory board member of the Scottish Lifesciences Association. He was recently appointed as a Fellow of the Royal Society of Edinburgh, UK.

Ken studied Electronics and Computer Science at Edinburgh University, Edinburgh, UK, and gained a PhD in image analysis and has four years postdoctoral research experience in medical image analysis. He returned to Edinburgh in August 2007 to join Canon Medical as R&D General Manager following his previous post of Operations Director for a European multinational where he was responsible for their imaging R&D facility in Cambridge, UK.

Canon Medical Research Europe
Canon Medical Research Europe works closely with global academic partners and clinical collaborators, as well as Canon Medical Group colleagues in Japan and the USA, towards translating state-of-the-art AI into effective clinical decision support that empowers clinicians and makes a positive contribution to healthcare.


Find out more about Canon Medical’s holistic approach to AI, powered by Altivity.