Copenhagen, Denmark – June 5, 2024 — Leveraging artificial intelligence (AI), breast radiologists in Denmark have achieved notable advancements in breast cancer screening, significantly enhancing accuracy while reducing false-positive findings. Published today in Radiology, a journal of the Radiological Society of North America, the study marks a pivotal moment in the integration of AI into medical diagnostics.
Mammography has long been effective in reducing breast cancer mortality but has been accompanied by the challenge of false-positive results. Recognizing the potential of AI to augment screening efficacy, researchers, led by Andreas D. Lauritzen, Ph.D., from the University of Copenhagen and Gentofte Hospital, embarked on a comprehensive study.
“We believe AI has immense potential to improve screening outcomes,” stated Dr. Lauritzen, emphasizing AI’s role in triaging normal results and supporting decision-making, thereby alleviating radiologist workload.
The retrospective study compared outcomes from two cohorts of women aged 50-69 undergoing biennial mammography screening in Denmark’s Capital Region. The first group, screened before AI implementation (October 2020-November 2021), involved dual radiologist readings. The second group (November 2021-October 2022) utilized AI for initial analysis, followed by either single-read or double-read interpretations.
The AI system, trained using deep learning models, identified suspicious lesions and calcifications within mammograms. Results showed that AI-assisted screenings detected more breast cancers (0.82% vs. 0.70%) with a lower false-positive rate (1.63% vs. 2.39%) compared to screenings without AI.
“In the AI-screened group, we observed a 20.5% decrease in recall rates and a 33.4% reduction in radiologists’ reading workload,” noted Dr. Lauritzen. The positive predictive value of AI screenings also surpassed traditional methods, detecting a higher proportion of smaller invasive cancers.
Despite these advancements, Dr. Lauritzen underscored the need for further research to assess long-term outcomes and guard against potential overdiagnosis.
As countries vary in breast cancer screening protocols, the study highlights Denmark’s pioneering efforts in integrating AI to enhance screening precision and efficiency. Future research aims to refine AI systems by incorporating patient history, further optimizing diagnostic accuracy.
For more information on the study’s findings and implications, the full report can be accessed in Radiology.