This week we spoke to Hyun Soo Ko, a frequent contributor to the ESR’s blog on artificial intelligence and a radiologist in the Department of Cancer Imaging at the Peter MacCallum Cancer Centre in Melbourne, Australia, as well as the Department of Diagnostic and Interventional Radiology at University Hospital Bonn in Bonn, Germany.
What is your background/experience with artificial intelligence and what first attracted you to the topic?
The earliest memory of computer-aided and AI concepts goes back to my primary school years while having conversations with my father, a now retired physicist and software engineer. He would describe mathematical/physics problem-solving from his job experience at a German industrial company. These case studies often revolved around methods to achieve the highest efficiency and how to make human tasks easier or more accurate/precise.
What are the biggest challenges to AI adoption in clinical practice?
AI needs to start from a place that humans can relate to – which is a complex task.
The successful integration of AI into clinical practice requires AI solutions to be impartial and transparent. They need to add value to clinical decision-making. In medicine, it takes time to establish trust with healthcare professionals as the main gatekeepers. Those who are dealing with AI advancements are acutely aware of their duty to stay up to date.
Any new AI solution must align with clinical logic to be seriously considered for adoption. It must embrace and balance personalized care as well as broad applicability, be founded on evidence-based medicine, and be adaptable across diverse healthcare settings.
Understanding these demands is crucial for AI to be impactful.
Give us an example (or an educated guess) of what you think AI will be able to do in 3 years. What about in 10 years?
I could well imagine that virtual friends will be more integrated in certain areas of our lives in the near future. An example could be having AI-driven virtual buddies within a virtual fitness or physiotherapy group session.
Predicting an AI achievement in 10 years is more difficult: Maybe a full-body acquired MRI scan that contains the complete array of common sequences including virtual contrast-enhanced images – obtained in 1 minute.
As a radiologist, what is your advice for younger colleagues wanting to dive into the topic of Artificial Intelligence, Machine/Deep Learning, and/or Radiomics? What are your tricks for staying up to date in this fast-evolving field?
Besides the more traditional approach of continuous learning and education, such as attending lectures and reading scientific journals, the internet and multimedia channels are excellent additional sources of information given their easy access, and often consist of shorter and more palatable content. However, one needs to be prudent to properly contextualize the presented and often abbreviated facts. It is important to be able to differentiate between fake and real news since AI is capable of hallucination and of creating incorrect content. AI has arrived and will keep evolving. Better to embrace and engage with AI now to stay ahead!
Recommended reading:
“The shift to personalised and population medicine”, J.A. Muir Gray
