A pilot lecture series at one medical center is designed to provide residents with in-depth instruction in commonly used artificial intelligence algorithms.
BI-RADS, LI-RADS, and Lung-RADS have helped countless radiologists classify breast, liver, and lung lesions. But, to date, little has been done to improve how providers understand the every-growing volume of artificial intelligence (AI) algorithms.
Dartmouth-Hitchcock Medical Center recently addressed this problem head-on by launching AI-RADS – an artificial intelligence curriculum for its radiology residents. Through a specifically designed set of lectures, residents not only reported they have a much greater understanding of AI, but they also indicated they were extremely satisfied with the program. Faculty shared these outcomes in an Oct. 17 article published in Academic Radiology.
“As artificial intelligence continues to reshape the world of medicine, it will become imperative that physicians are familiar with fundamental algorithms and technique in artificial intelligence,” said the team led by Petra J. Lewis, MBBS, professor of radiology and vice chair of radiology education. “This will become an essential skill for interpreting medical literature, assessing potential clinical software augmentation, formulating research questions, and purchasing equipment.”
The field of AI in radiology is constantly shifting, Lewis’s team said, and the level of instruction radiologists have received in how to understand and implement these tools has been uneven at best. In their pilot course, they designed a series of lectures that offer guidance on the strengths and limitations of machine learning with the goal of helping providers better understand the scientific literature being published on the subject.
For more coverage based on industry expert insights and research, subscribe to the Diagnostic Imaging e-Newsletter here.
To maximize AI understanding, the team crafted each of seven lectures around specific algorithms and supporting fundamental concepts in computer science. The team selected which algorithms to include based upon how commonly they are used and the problems they were designed to solve, using them to build an intellectual foundation for data representation. Lectures were held monthly for seven months.
Simultaneously, residents participated in a journal club where participants discussed each lecture’s algorithm. These two-hour meetings were held within two weeks of each lecture and included study guides that defined unfamiliar terms and broke down complicated mathematical expressions into simpler terms.
Overall, the team said, after attending the lecture series, residents were expected to be able to:
The team surveyed residents with content questions to assess how much their understanding of AI improved after the lectures, and they also asked participants to rate their satisfaction with lecture series. Exit surveys indicated a high level of happiness – 9.8 out of 10.
“Trainees overwhelmingly feel that the content depth of the AI-RADS lecture series is ideal, and the examples used are helpful vehicles to understand the key concepts in artificial intelligence,” the team said.
Although the pilot program was implemented as an in-person curriculum, the team said there are plans underway to transition it to an online curriculum. In this way, they said, it can better accommodate the scheduling needs of a wider group of residents. In fact, they said, videos will be uploaded to the medical center’s YouTube channel as they become available.
Ultimately, the team said, their goal is to ensure the highest level of training for the next generation of radiologists.
“Residency programs are only beginning to employ basic computing concepts in their training, a skill that will become essential for the radiologists of tomorrow,” they said. “Proficiency in artificial intelligence will be a required skill in the near future of imaging services.”