The United States published more AI-based radiology studies than any other country across the globe, according to a new analysis released Aug. 15 in the American Journal of Roentgenology.
In fact, the U.S. is responsible for nearly half of all total radiology AI studies, wrote Elizabeth West, with Columbia University Medical Center’s New York Presbyterian Hospital, and colleagues.
“There has been an exponential growth worldwide in AI radiology research, led by the United States, which produces almost half of global AI research,” the authors added. “With such strong support for AI radiology research, the United States is home to the three institutions of highest publication productivity.”
The team searched a comprehensive central database using a number of radiology-specific AI search terms such as: “AI,” “CNN,” and “neural network” published between 2000 and 2018. A statistical software program performed linear and nonlinear regression analyses to compare different country’s publishing outputs. In total, the analyses produced 8,813 publications.
Of the top 12 most productive countries, the U.S. was by far the leader in publications, accounting for about 35 to 50% of total AI-based radiology research. China, Germany, the United Kingdom, Canada and Japan were also among the top publishing countries. The authors pointed out that China contributed 18% of the total output in 2018, making it a growing competitor to the U.S.
“China's ability to exponentially increase productivity is likely due to the country's unique research infrastructure,” the team wrote. “The availability of large centralized data and rapid implementation across commercial industries have already helped the nation become very productive in AI research in a short period. In addition, Chinese government directives and funding for the advancement of AI have generated an incredible mobilization in research and development among Chinese researchers.”
Studies need money to be published. The authors found the top countries that funded AI publications were the U.S. (16.5%), China (3.6%), the UK (1.5%) and Canada (0.6%). In the U.S., the National Institutes of Health, National Cancer Institute and National Science Foundation poured the most money into AI research.
Broken down by subspecialty, neuroradiology (24.4%) led the pack. Chest and body (20.7%), nuclear medicine (13%) and breast imaging (12%) followed close behind.
“This is not surprising, given that neuroradiology is a unique subspecialty of acuity, where ‘time is brain’ for stroke evaluation demands fast and accurate diagnoses suitable for AI applications,” West and colleagues wrote.
Limitations of the study included a “likely underestimation” of total AI research productivity and the fact that the analysis did not account for the quality of publications.
Going forward, the researchers believe the U.S can be successful if the government continues to prioritize and fund AI radiology research.
Matt joined Chicago’s TriMed team in 2018 covering all areas of health imaging after two years reporting on the hospital field. He holds a bachelor’s in English from UIC, and enjoys a good cup of coffee and an interesting documentary.