Bringing the applications of Artificial Intelligence (AI) into the field of radiology is a significant ongoing priority of the Canadian Association of Radiologists (CAR), and CAR member Dr. Alexandre Cadrin-Chênevert is working to realize the possibilities of the ever-evolving technology.
Dr. Cadrin-Chênevert is an Associate Editor with the Radiological Society of North America’s (RSNA) journal Radiology: Artificial Intelligence. He says that finding AI applications that enhance radiological workflows and improve patient outcomes is crucial work for the present and future of medical imaging.
“The development and safe clinical integration of artificial intelligence algorithms is a critical issue for the future of radiology, both in terms of quality and accessibility over a 20-year horizon, and the active participation of radiologists in the process is paramount for a smooth and beneficial technological evolution for all.”
Dr. Cadrin-Chênevert says the journal has developed momentum rapidly and is becoming an international scientific leader in this field of research, which has grown in recent years.
“Deep learning research applied to medical imaging has literally exploded since 2017 and is now a significant proportion of research in all radiology subspecialties. In a rapidly changing environment, involvement in the Editorial Board helps keep up with emerging trends in the literature.”
Multi-institutional deidentified data sharing for training/validation is the key to foster generalizable innovation.
Even more important for low-prevalence cancers.
When data sharing is not legally or ethically possible, federated learning is a decent alternative.#RadAIchat https://t.co/rF4Bm6Rozi
— Alexandre Cadrin-Chênevert (@alexandrecadrin) September 8, 2022
In order to advance this complex field of research, it is necessary to have a breadth of knowledge from experts in different areas, says Dr. Cadrin-Chênevert.
“The diversity of board members from the fields of Computer Science, Machine Learning, and Radiology provides a unique forum to foster mutual learning and fluid communication among these experts.”
While there are innumerable paths of inquiry in AI research, Dr. Cadrin-Chênevert has a specific area of research interest.
“I am significantly interested in the public sharing of de-identified data and the creation of large medical imaging datasets. In my opinion, this is the single most important factor in this field of research to maximize the scientific impact and to promote a fair, safe and reproducible development of these algorithms,” he said. “In this context, I recently wrote a commentary in our journal presenting an article where the authors have shared publicly a large dataset of more than 1.35 million images named RadImageNet which will undoubtedly have a marked effect on this field of research.”
Looking ahead, Dr. Cadrin-Chênevert is excited for the possibilities of synergizing AI tools with the work of radiology and believes the results to come will be desperately needed for the maintenance of our national healthcare system.
“In Canada, our aging population will continue to put a lot of pressure on our healthcare system in the near future, with a significant increase in demand for radiology services in a context of scarce resources. As AI tools become more mature, robust, and secure, their clinical integration will increase our productivity to promote the quality and accessibility of our services. In our context, it won’t just be exciting, it will be necessary.”
As the field of AI research moves forward, Dr. Cadrin-Chênevert believes it is important for the journal to maintain an academic presence on social media. Interested readers can engage with the Radiology: Artificial Intelligence by following @Radiology_AI on Twitter and can find their latest publication online.