Using artificial intelligence (AI) to support mammography screening has been found to be a safe and effective alternative to traditional double reading by radiologists, according to a trial led by researchers from Lund University in Sweden. The interim analysis of the trial, published in The Lancet Oncology, revealed that AI-supported screening boosted cancer detection by 20% and reduced radiologist workload by 44%.
Mammography screening plays a crucial role in detecting breast cancer, and every year around one million women in Sweden undergo this examination. Currently, each screening is reviewed by two breast radiologists in a process known as double reading. However, a shortage of breast radiologists poses a risk to the screening service. AI has been increasingly explored as a potential solution, but its optimal implementation and clinical consequences remain unclear.
To better understand the impact of AI in mammography screening, the Mammography Screening with Artificial Intelligence (MASAI) trial was designed as the first randomized controlled trial to evaluate the effects of AI-supported screening. The trial randomly allocated 80,033 women into two groups: 40,003 women in the intervention group underwent AI-supported screening, while 40,030 women in the control group underwent standard double reading without AI support.
Using AI resulted in the detection of 20% more cancers compared to standard screening, without increasing the rate of false positives. False positives occur when a woman is recalled but subsequently cleared of cancer suspicion. Moreover, the workload for radiologists was significantly reduced, with 46,345 screen readings performed with AI support compared to 83,231 with standard screening. This reduction in workload saved approximately five months of a radiologist’s time to read the same number of screening examinations.
The study was conducted at a single site in Sweden, and further research is needed to validate these results under different conditions and with other AI algorithms. The MASAI trial aims to enroll a total of 100,000 women to assess interval-cancer rate, which refers to cancers diagnosed between screenings and generally carries a poorer prognosis than screen-detected cancers. The research team also plans to investigate the types of cancers detected with and without AI support.
Screening is a complex process, requiring a careful balance between benefit and harm. The objective is to identify clinically significant cancers at an early stage while minimizing false positives and overdiagnosis of indolent cancers. The initial analysis of the MASAI trial demonstrates that AI-supported screening is safe, as the cancer detection rate did not decline despite the substantial reduction in radiologist workload. The forthcoming analysis of interval cancers will provide further insights into the accuracy and efficacy of AI-supported screening.
In conclusion, the use of AI in mammography screening shows great promise in improving cancer detection rates while alleviating the heavy workload on radiologists. As the MASAI trial progresses and more data is collected, a clearer picture will emerge regarding the long-term benefits and potential implications of AI-supported screening. Nonetheless, this exciting advancement offers hope for a more efficient and effective approach to breast cancer screening.
