Autonomous AI reduces radiologists' breast cancer screening workload in real-world setting
A new prospective analysis is offering a renewed sense of optimism for how artificial intelligence support can reduce the workloads of radiologists in real-world clinical settings.
Published in Nature Medicine, the study details how a partially autonomous AI tool deployed in screening mammography identifies normal, low-risk exams and eliminates the need for double reads. During the trial, the tool effectively reduced readers’ workloads while also increasing cancer detection rates.
The prospective nature of the study signals that the tool can operate safely in clinical settings. While there are numerous studies on using AI to increase lesion detection and decrease interpretation burdens, it is important to determine how these tools perform in real-world settings in real-time. Whether these tools can function autonomously is an even bigger question—one that these latest findings might help begin to answer.
For the study, researchers deployed two reading strategies for the exams of over 30,000 women who underwent breast cancer screening in Spain between 2022 and January 2024. The first strategy entailed standard double-blind reading and the second incorporated partially autonomous AI-supported reading, where cases classified as low-risk were deemed normal and the rest were double read with AI support. The team analyzed how each method impacted radiologists’ workload, cancer detection and recall rates.
With AI’s support weeding out the normal exams, radiologists’ workloads were reduced by more than 60%. At the same time, cancer detection rates increased by over 15%, while recall rates for AI as a first reader increased by around 14%. These results were consistent across both digital breast tomosynthesis and digital mammography.
The results are promising and could represent a real opportunity to reduce reading burdens for radiologists, the team noted. However, they acknowledged that deploying autonomous AI into clinical practice is attached to many legal and ethical considerations.
“The ethical implications of relying on AI for initial reads versus radiologists must be carefully considered to ensure patient safety and trust in screening processes,” Esperanza Elías-Cabot, MD, of the Reina Sofía University Hospital and the University of Córdoba in Spain, and colleagues cautioned. “If there is no radiologist review of a significant proportion of exams, additional screening quality assurance processes such as automated mammography image quality control and continuous postmarket surveillance of AI performance are necessary steps before implementing such an autonomous AI screening workflow for the very likely normal screening exams.”
Read more from the team’s research here.
In addition to her background in journalism, Hannah also has patient-facing experience in clinical settings, having spent more than 12 years working as a registered rad tech. She began covering the medical imaging industry for Innovate Healthcare in 2021.