How Artificial Intelligence is Revolutionizing Radiology

The number of life science papers describing artificial intelligence (AI) rose from 596 in 2010 to 12,422 in 2019 —an increase of more than 2,084 percent in only ten years. According to one peer-reviewed article published in Nature, “AI shows potential for diagnosing, managing and treating a wide variety of medical conditions.”

Nowhere is this potential more apparent than in radiology.

Over the last decade, peer-reviewed publications on AI for medical imaging have increased from approximately 100-150 per year to 700-800 per year. Magnetic resonance imaging and computed tomography account for about half of these publications; neuroradiology, roughly a third. However, experts predict that AI will soon revolutionize every type of radiology, with applications for image acquisition, processing, interpretation, reporting, and more. [2] 

In 2020, the United States Food and Drug Administration approved 100 AI-enabled medical devices. In 2021, it published a 350-item list of approved AI medical devices, 70 percent for medical imaging. [3] 

The Key Differentiators Driving Growth

Radiology’s natural abundance of visual data makes it well-suited to the early application of data-driven technologies like AI. Moreover, AI offers convenient solutions to two of the industry’s top challenges:

#1 - Precision

Reading medical images can be like searching for a need in a haystack. Nodules, lesions, and other health indicators targeted by medical imaging can be small and subtle—sometimes even invisible to the human eye.

AI can help radiologists read medical images by augmenting their human expertise with the accuracy of machine vision and the power of computer processing. A study published in Clinical Radiology found that AI helped radiologists reduce the rate of missed lung cancers in chest X-rays by 60 percent.[4] Another found that AI could help radiologists identify microscopic fractures invisible to the human eye with a 90 percent success rate.[5]

#2 - Workload

Experts estimate a shortage of 9.9 million healthcare professionals by 2030. This labor crisis is already in effect; one study found that the mean on-call workload for radiologists more than quadrupled between 2006 and 2020.

AI can help radiologists do more work in less time, reducing their workload (and burnout) by automating tasks and, where automation is impossible, increasing efficiency. A paper published in Pediatric Radiology suggests that “computer-aided detection (CAD) can decrease reading time by making the diagnostic process easier” and “[AI-enabled] image enhancement could not only shorten image acquisition time but also ease detection.”

Looking to the Future

Publication, approval, and adoption rates have risen dramatically over the last decade, to the benefit of patients and practitioners. Moreover, experts forecast that this growth will not slow. AI will continue to increase diagnostic accuracy and precision, reduce workload, and otherwise optimize radiologists’ work across specialties.