Research results from the Mammography Screening with Artificial Intelligence (MASAI) study at Lund University show that breast cancer screening using AI detects 29 percent more cases than traditional screening. Also, significantly more invasive cancers were detected at an early stage using AI. The last part of the study focused on breast cancer that is missed by screening.
The first results of the MASAI study - a randomized study to evaluate whether AI can improve mammography screening - were already published in August 2023. The study started in the spring of 2021 and the final report is expected in 2026. The new research report is based on the results of almost 106,000 women who were screened for breast cancer. Half of them were randomly assigned to traditional mammography screening, while the other half received AI-supported screening. AI-supported screening not only detected more cancers overall (338 people compared to 262), but also 24 percent more invasive early-stage cancers (270 people compared to 217).
Strong numbers
Kristina Lång, who is responsible for the study, is pleased to be able to show strong numbers. According to her, the research also shows that relatively more aggressive cancers are detected, which are very important to detect early. “If it is detected at a later stage, the chances of survival are worse and more intensive treatment is often needed,” says Lång. She also points out that people suffer less and treatment costs are lower if the cancers are detected at an early stage.
Every year, approximately 1 million women in Sweden are invited for a mammography screening. The mammograms are assessed by two breast radiologists. And there is currently a shortage of these radiologists. This is one of the reasons why the MASAI study focused on using AI to identify mammograms with an increased risk of breast cancer. In these cases, the radiologist was assisted by AI, which flagged suspicious findings on the images. As in the previous report, AI-supported screening also showed that the workload of breast radiologists was reduced by 44 percent. The new findings were recently published in The Lancet Digital Health.
Implementation of AI support
Sweden has a generous screening program by international standards. All women between the ages of 40 and 74 are invited for a mammogram every 1.5 to 2 years. However, the interval between two screening visits can be long enough to detect cancer, even if the last screening was considered normal. These so-called interval cancers are the next group of cancers that Lång and her colleagues will analyze. The results of the MASAI study have now contributed to the implementation of AI support in several regional screening programs in Sweden.
Earlier this year, an observational study in Germany tested an AI-supported method for analyzing mammograms of women between the ages of 50 and 69. That study showed that using AI, in which the mammograms are screened twice, increased the percentage of breast cancer diagnoses by 17.6 percent.
