Thanks to data compiled by the U.S. Food and Drug Administration (FDA), we know the agency has cleared 521 artificial intelligence and machine learning (AI/ML)-enabled medical devices, with some approvals going back decades. The pace of such approvals has accelerated. The first device listed had its authorization date in 1995. The vast majority of the authorized devices are in the radiology space.
An article just published by Modern Healthcare offers a good summary of the state-of-the-art for adoption of AI in radiology. The latest polling by the American College of Radiology indicates that nearly one-third of radiologists are using AI in imaging and 20 percent of practices said they plan to invest in AI tools in the next five years, But today, and on the threshold of a new year, that data is two years old. It feels like far more AI integration than the 2021 ACR survey reports.
Modern Healthcare notes that AI utilization is significant in breast cancer detection. Also, newer photon-counting detector CT technology paired with AI-based noise reduction can process CT images quicker with better resolution and reduce the amount of time needed for CT scans, lowering patients' radiation exposure.
Challenges for AI improvement will be to turn focus to continuous machine learning by automatically retraining AI models with new data on a regular basis, and to access healthcare’s historically siloed data sets, which are not interoperable.
One thing that is clear is that bright minds are working diligently to find new and improved uses for AI in radiology. It will be fascinating to see how AI uses in imaging mature in 2023 and beyond. What will we be writing about at the threshold of 2024?