Reshaping the future of AI education in radiology

 

Artificial Intelligence (AI) is revolutionising the field of radiology, making faster and more precise diagnoses possible. However, most radiologists aren’t familiar with AI and report a fear of being replaced by it.

 

Dr Jordan Perchik from the University of Alabama at Birmingham, USA, designed a free online ‘AI Literacy Course’ which has reached more than 500 radiology trainees from 10 countries.

 

Read more in Research Features

 

Read the original research: doi.org/10.7191/jgr.783

 

Image Credit: Adobe Stock / Primopiano

 

 

Transcript:

Hello and welcome to Research Pod! Thank you for listening and joining us today.

 

In this episode we look at the work of Jordan Perchik from the University of Alabama at Birmingham, USA, who, has designed a free online ‘AI Literacy Course’ to provide more than 500 radiology trainees from 10 countries with revolutionising medical training. While artificial intelligence, or AI, has been revolutionising the medical field, most radiologists themselves are not familiar with AI, and many report a fear of being replaced by it.

 

Artificial intelligence is a fast-growing area of computer science that is making a significant impact on medicine, especially in the field of radiology. AI algorithms can be unique tools to help radiologists reach a faster and more precise diagnosis. Despite the obvious advantages, many radiologists remain reluctant to incorporate AI into their domain, while the majority of them have limited knowledge of AI applications. Dr Jordan Perchik from the University of Alabama at Birmingham, USA and his coworkers designed a free online course packed with the essential tools for AI use in radiology, targeting radiology residents as well as medical students. This course ran at an international level and managed to increase understanding of AI, while also proving to be an excellent tool for encouraging radiologists to familiarise themselves with AI.

 

In recent years, AI has seen a dramatic increase in the field of radiology, while AI applications remain at the centre of healthcare. In essence, AI can be described as a set of tools that are programmed to perform different kinds of tasks that typically would be considered to require human intelligence and reasoning. Among many other applications, AI aims to solve practical problems in healthcare, particularly in radiology. Radiology is the medical domain that specialises in using imaging technology such as X-ray, CT, MRI, and ultrasound to diagnose and treat diseases. What AI can do, among other tasks, is assist the radiologist by automatically detecting complex patterns in the images and assist in diagnosis, thus reducing the diagnosis burden and shortening the diagnosis time. Despite the possible benefits, radiologists around the world lack appropriate AI training; additionally, they seem very reluctant for AI to enter their field with some fearing that AI will replace the human radiologist.

 

As AI becomes the centre of attention by being a common subject in a plethora of conferences and publications, incorporating AI education into the curriculum still seems to be challenging. Few AI courses are available for radiologists looking to learn more about AI, and these courses are often expensive and require an extensive time commitment. Additionally, these courses often spend a substantial amount of time covering materials not relevant to the clinical use of AI in radiology.

 

These barriers to AI education access are even more pronounced for radiologists in low- and middle-income countries. As a consequence, radiologists globally report they lack training in AI. But is this trend irreversible? And if so, how is it possible to provide AI education on a global level? Perchik and colleagues believe that a free and online training course could become the catalyst needed for the safe adoption of AI applications in radiology.

 

Indeed, studies have shown that radiologists who have received some training on AI seem to value the possible AI applications in their everyday work. That is why Perchik and colleagues designed ‘The AI Literacy Course’, led by the Artificial Intelligence in Radiology Education group at the University of Alabama at Birmingham since 2020. In 2021, the week-long online course included nine training programmes and reached over 150 radiology trainees. Specifically, the participants targeted were from training programmes in the Southeast and Mid-Atlantic United States. The curriculum for the AI Literacy Course curriculum was developed by a panel of three attending radiologists and a lead radiology resident, and contained core subjects such as Algorithm Bias and Ethics of AI. The course was incredibly successful and managed to improve familiarity and comfort with AI terminology and applications.

 

In 2022, Perchik and colleagues further expanded the AI Literacy Course by addressing participants on an international level, from a total of 25 residency programmes. Participants from 10 countries registered for the course, including the US, Colombia, Grenada, the Netherlands, Cameroon, Nigeria, Egypt, Lebanon, Saudi Arabia, and India. The course covered a wide range of topics, from introductory lectures on AI to lectures on applications of AI in the subspecialties of Nuclear Medicine, Paediatric Imaging and Musculoskeletal Radiology and finally, special topics lectures, such as ‘The future of AI radiology’ and ‘Federated learning in AI’.

 

A hands-on session was also included in the 2022 curriculum, giving the participants the opportunity to familiarise themselves with an FDA-approved AI tool for advanced cancer. In addition, the courses were recorded and uploaded on YouTube, giving participants the opportunity to watch the curriculum in their own time. In fact, the course managed to reach more than 500 participants, which is more than threefold higher than the number of participants reached in 2021.

 

The week-long course for radiology trainees was held from 3–7 October 2022. The course was evaluated by handing out a survey before and after the completion of the course. Before the course, 64.2% of the participants reported a low familiarity with fundamental terms, methods, and applications of AI in radiology. Impressively, at the end of the course, almost all participants, that is 93.2%, reported that the course increased their understanding of AI, and 86.4% reported interest in participating in radiology AI courses in the future. Interestingly, 59.1% reported that this course was their first chance to cover AI in radiology.

 

These spectacular results can be attributed to the general format of the course, such as the quality of the curriculum which covered from fundamental AI topics to more in-depth AI applications in radiology, as well as the hands-on session which helped the participants to familiarise themselves with an AI tool. In particular, it is important to mention that the remote element of this course played a major role in its success, as this gave participants the opportunity to join, no matter their location. Additionally, since all lectures were recorded, the participants were able to follow the course at their own pace.

 

The AI Literacy Course proved to be an effective tool for increasing accessibility to AI education for radiology trainees around the world. These remote education seminars managed to reach out to different participants, from radiology trainees to practicing radiologists to medical students, and served as a free source of practical and easily accessible AI education. The Artificial Intelligence in Radiology Education group aims to repeat the free courses on an annual basis, using the same remote education format, with the ultimate goal to keep AI education accessible and help radiologists identify the grand benefit they could have by using AI in their practice.

 

Speaking of the future of AI, radiology, and education, Dr Jordan Perchik remarks that:

 

‘In future courses, we are hoping to add more opportunities for hands-on experiences. We want participants to be able to see AI in action, whether this is with a demonstration of a current commercial AI product or training an algorithm themselves on a pre-prepared set of radiology images.’

 

It has been shown that fear of AI goes down when experience with AI goes up. Much of the fear of AI in radiology comes from uncertainty of how AI will affect the radiologist’s role in healthcare. For many of our participants, this course is their first experience with AI in radiology. Our goal is that after our participants see what AI can do, and just as importantly, what it can’t do, they will be more welcoming of AI integration into the practice of radiology.’

 

‘We hope that our course will make AI education more accessible to the radiology community. In the near future, when AI is more widely clinically integrated and concepts of AI are tested on radiology licensing exams, it will be necessary for programmes to have their own AI curricula and their own AI experts on site. This is not the reality we are in currently, and many programmes in the US and internationally do not have these resources in place. In the meantime, we hope our course can serve as a bridge to provide free and accessible AI education to radiologists in training and to also serve as a model for programmes looking to start their own programmes in the future.’

 

That’s all for this episode – thanks for listening, and stay subscribed to Research Pod for more of the latest science.

 

See you again soon.

Leave a Reply

Your email address will not be published.

Top
Researchpod Let's Talk

Share This

Copy Link to Clipboard

Copy