AI Accurately Detected 20% More Breast Cancers Than Human Radiologists, Study Finds
A recent study has demonstrated that a combination of a trained radiologist and artificial intelligence (AI) can improve breast cancer detection in mammogram screenings by 20% without increasing false positives.
The study placed two trained radiologists against the man-machine team, which the authors claim is the first randomized controlled trial showcasing the efficacy of AI in breast cancer screening.
The study, which was conducted in Sweden and involved 80,000 mammogram screenings, found that the human-AI collaboration successfully identified breast cancer in 6 out of 1,000 women as compared to the rate of 5 in 1,000, which was the count of the two radiologists.
Incredibly, this 20% improvement in detection didn’t lead to an unusually high rate of false positives either, proving the AI’s balanced sensitivity.
Although the integration of AI into mammogram screenings won’t result in robots taking over anytime soon, the study does signify that AI-assisted detection can offer a safe and effective means of enhancing the speed, efficiency, and accuracy of breast cancer screenings.
Study co-author and associate professor of radiology diagnostics from Lund University in Sweden, Dr. Kristina Lång, told CNN, “The greatest potential of AI right now is that it could allow radiologists to be less burdened by the excessive amount of reading.”
Dr. Lång mentions that one particular outcome of the study was the 44% reduction in the workload of the human radiologist within the human-AI partnership. This reduction in workload could alleviate the burden on radiologists who are already facing increasing demands due to the rising average age of the Western population and the growing prevalence of breast cancer.
Studies prove that early detection of breast cancer significantly improves survival rates, making routine mammograms crucial for aging women. As cancer rates continue to rise over the years, it’s palpable that the workload of radiologists will only increase, underscoring the potential benefits of AI assistance.
Previous reports have shown AI’s ability to identify abnormalities in chest X-rays just as effectively as trained radiologists with a sensitivity rate of 99.1% . This further highlights the potential of AI in the medical diagnostics field.
The study’s editorial praised AI’s capacity to handle a substantial portion of normal readings, around 7.8%, relieving radiologists of these tasks and positioning AI as a labor-saving tool rather than a threat to job security, as shared by CNN.
In line with this, radiologists themselves have emphasized that AI serves as a valuable time-saving tool and does not jeopardize their job security. However, it is universally agreed that AI, if intended as a labor-saving innovation rather than a life-saving one, must undergo rigorous testing to ensure it does not compromise patient safety for the sake of reducing radiology department workloads.