AI triage tool slashes breast cancer screening workloads by 77%
New data detail the significant workload reductions artificial intelligence-enabled mammogram triage could offer breast radiologists.
Members of the subspecialty look at similar images for hours on end every day. This redundancy can lead to missed findings, which is why many institutions double-read mammograms. Though effective, this practice represents a substantial workload for radiologists, making it a prime target for AI optimization.
There is ample research backing the technology’s ability to detect abnormal findings and stratify risk. Updated guidelines from the National Comprehensive Cancer Network reflect these proven benefits, as the NCCN now recommends image-based AI risk assessment as a primary tool for identifying increased risk of breast cancer.
Researchers recently deployed AI as a first reader in a double reading setting to assess its impact on radiologist workloads, with the workflow only requiring negative exams to undergo a second interpretation. The tool was retrospectively applied to 55,589 screening mammograms from 42,419 women age 50–74 in a French screening cohort, with outputs compared to previously filed reads. Researchers calculated how the algorithm’s concordant interpretations could have reduced workloads for initial interpretations.
AI produced 183 additional recalls, 12 of which eventually resulted in a cancer diagnosis. It classified 76.6% of exams as low risk and 23.3% as non-low risk. Just one of the additional cancers came from the group AI deemed as low-risk.
“AI triage could potentially reduce second-reading workload by approximately 77% in the French breast cancer screening program, despite a small but measurable risk of missed cancers and the need for governance,” Benoît Mesurolle, MD, from the Republic Radiology Center in Clermont-Ferrand, France, and colleagues commented.
The group signaled support for the use of such AI tools in breast cancer screening, which would allow radiologists to focus on higher-risk cases. Additional, prospective evaluations of AI performance in these settings, however, are still warranted, they added.
Read more 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.