The leadership of the ACR has made connecting with and listening to members-in-training, including medical students, a priority — with the goal of keeping the pipeline to radiology open to the best and brightest. Students, trainees and rising radiologists have their own unique outlook on the future of the specialty, and their thought processes and contributions to the field are vital as the College develops and implements its strategic plan. To advance ACR’s commitment to better engage medical students and provide more volunteer opportunities, the Medical Student Subcommittee has been elevated to the Medical Student Section (MSS). The new Section reflects the value medical student members bring to the College’s entire membership.

Comprised of ACR members-in-training, the MSS currently represents more than 3,600 medical students and is led by a nationally representative steering committee of medical students and a Resident and Fellow Section (RFS) advisor. The goal of the MSS is to develop resources to benefit students who are exploring diagnostic or interventional radiology, nuclear medicine, radiation oncology or MRI physics as future careers. This is the first article in a new Bulletin series proposed and authored by medical students — sharing their perspectives on challenges, opportunities and outcomes within an ever-evolving specialty.

AI encompasses various technologies, such as machine learning, natural language processing and deep learning. AI works in tandem with computer systems to carry out tasks that typically require human intelligence, such as visual perception, speech recognition and decision-making. In the past seven years, AI in oncology has made considerable progress in interpreting sensory information and representing complex data in both radiology and non-radiology fields. High-performance computer-aided detection (CADe) has shown similar performance compared to human readers in mammography, and digital pathology image analysis software is able to accurately perform mitosis detection, segment histologic primitives (such as nuclei, tubules and epithelium) and both characterize and classify tissue.

Instead of considering AI’s potential to enhance the value of medical imaging, misguided learners may clutch their pearls in the face of AI’s success in image recognition tasks.

Numerous fields within healthcare are being transformed and augmented by the growing capabilities of AI. No other medical field has been a focal point for this shift more than radiology. In 2016, Geoffrey Hinton — who pivoted his career to focus on the perils of AI — famously stated that there was no further need to train radiologists, asserting that machine learning would surpass the skills of radiologists in the near future. Sensationalized reports of AI outperforming diagnostic radiologists in detecting pneumonia soon followed, and the topic was even touched on in the 2019 presidential primary debate.

As the growth of AI in medicine continues to increase, a dichotomy has emerged between radiologists and non-radiologists. Notably, there is an over-representation of exaggerated concerns in medical trainees (as well as non-stakeholders in the field of radiology) compared to seasoned radiologists and radiology trainees. For example, 88% of a surveyed cohort of UK medical students believed AI will have a profound impact on the field of radiology. Among those surveyed, almost half reported to be less likely to consider a career in radiology due to the perceived success (and possible threat) of AI.

An analogous scenario to the current AI “bubble” occurred during the roll out of the Picture Archiving and Communication System (PACS) more than four decades ago. This revolutionary technological innovation transformed the field of radiology and made imaging more accessible to those in the healthcare field. Some radiologists worried that PACS would diminish their role and even lead to a reduction in the demand for their expertise. After all, once everyone could view digital images, bypassing radiologists and self-interpreting images could easily follow. What followed instead was the monumental integration of PACS into the fabric of 21st-century medicine. PACS has improved workflow capabilities, enhanced communication between providers and shortened the average time to diagnosis.

Central to the concerns of early trainees, how the radiology profession discusses the impact of AI on their work can shape how the greater medical community understands the capabilities of AI applications. In an interview with ASNR Pearls, Jody Tanabe, MD, chief of neuroradiology at the University of Colorado, emphasized the critical role radiology plays throughout the healthcare system. Radiology is not merely a service, Tanabe says, but an integral component in the delivery of healthcare services. “We’ll never get replaced by AI,” she predicts. Looking through the lens of how brain imaging has evolved in the past 20 years — particularly in the diagnosis and management of stroke — there will consistently be new discoveries to sustain radiology as an invaluable asset. In Tanabe’s words, “I welcome the day that AI will find the incidentals so I can focus on the interesting questions.”

Just as pilots rely on avionics to manage complex data and maintain safety, radiologists can use AI to manage tedious, mundane tasks to assist in diagnosis. Detection algorithms will identify issues like breast calcifications and lung nodules, while measurement apps and classification routines streamline other processes. This allows radiologists to focus on higher-level cognitive functions, exercising judgment, creativity and empathy.

A critical step forward in the field of radiology must include how we discuss the implementation of AI into the specialty. This must be tempered by the reality that many students and trainees harbor misguided fears about machines replacing the field as a whole. Medical students have a limited understanding of AI, with many reporting that they learned about it more from mainstream media than from their university courses. Addressing these preconceived notions may alleviate apprehensions about the future of the field. Additionally, it is important to note that students who received an education about AI were less likely to rule out radiology as a career choice. To echo the words of Shreyas Vasanawala, MD, PhD, professor of radiology at Stanford, we must view AI as assistive, not adversarial.

Medical students interested in submitting an article for publication in the ACR Bulletin can sign up via Volunteer Link. You may contact the authors at cbolles@vcom.edu and Dallinjudd@my.unthsc.edu, respectively.