Radiology professionals working on artificial intelligence (AI) technologies can learn a lot from studying self-driving vehicles, according to a new commentary published in the Journal of the American College of Radiology.
Lead author Kimberly Powell works for Nvidia, a Santa Clara, California-based technology company. Powell opened the commentary by sharing her own story. Radiologists from the Johns Hopkins University School of Medicine in Baltimore then served as co-authors as she explored AI’s potential impact on radiology.
Powell wrote that she has watched Nvidia’s growth over the years from a gaming company to AI specialists. Nvidia now plays “an important role in making self-driving vehicles a reality,” she explained.
“Among the unprecedented innovations made possible by AI is the self-driving vehicle,” she wrote. “Self-driving cars have the potential to deliver improvements in safety, efficiency, costs, and business prosperity. Traffic accidents and the resulting fatalities will no longer occur. Daily commutes will be transformed into time we can spend on other tasks.”
Nvidia developed its Drive AI platform during its research, Powell wrote, and the company had to make sure it always learned from any mistakes along the way.
So what can the imaging industry learn from Powell’s experience? Quite a lot, it turns out.
For example, Powell and colleagues noted that rapid progress is being made right now in AI. Radiologists might view AI as something that is many years down the road, but like self-driving vehicles, things are going to change faster than a lot of people realize. “It is important to realize that many of these features are not far-future applications in radiology but will be incorporated into routine clinical practice over the next few years,” the authors wrote.
In addition, radiologists can view the impact of AI on their own work similar to how Nvidia views the impact of self-driving vehicles on commuters.
“Instead of thinking about AI replacing our abilities, we should instead view AI as allowing us to ‘level up’ what we can accomplish,” the authors wrote. “In radiology, having AI perform the mundane tasks that humans may struggle with or find interminable, such as pulmonary nodule detection or organ segmentation, as well as the tasks that are simply impossible for humans, such as extracting lesion textural features across multiple contrast phases or sequences, frees us to interact with the images in ways that can push the boundaries of diagnostic science. By using AI, we can add depth to our readings that can produce tangible benefits to patient care.”
Michael has more than 16 years of experience as a professional writer and editor. He has written at length about cardiology, radiology, artificial intelligence and other key healthcare topics.