To detect breast cancer early on, women who are at a high risk of developing the disease can now be recognized during mammography screening exams, thanks to artificial intelligence.
Image Credit: HealthyCapture Studio/Shutterstock.com
Now, a multinational study team under the direction of Sweden’s Karolinska Institutet can demonstrate that the technique works across several European nations. The Lancet Regional Health - Europe published the research.
To improve the likelihood of early detection, women with a high risk of breast cancer may benefit from additional exams, according to an AI-based risk model that analyzes mammography pictures. Having tested the approach on more than 8,500 women in Germany, Spain, and Italy, the researchers are now able to demonstrate that the model performs well across various demographics.
Women are checked for mammograms under current programs in a set age range (40–74 in Sweden) and interval, typically every other year. Nonetheless, studies have indicated that individual risk factors for breast cancer differ, meaning that women would gain more insight into their own risk with personalized screening. Risk models have been around for a while and are frequently predicated on lifestyle choices and a woman’s family history of breast cancer.
AI Detects Tiny Changes
Researchers have created a whole new class of risk models based on minuscule alterations in the photos that are too small for the human eye to notice by allowing a trained AI to analyze screening photographs.
It’s not as simple as traditional models that use a handful of factors such as genes, as there are thousands of factors in the image that are taken into account. The AI is able to find different patterns in these factors, each of which are weak but that the AI can combine. The AI can also give an overall assessment of what is likely to happen in the breast in the future.
Mikael Eriksson, Study Leader and Postdoctoral Researcher, Department of Medical Epidemiology and Biostatistics, Karolinska Institutet
Many women today receive late-stage diagnoses, and breast cancer can even develop in between screenings. To detect any tumors early on, the AI-based risk model can be used to identify which women require further testing in addition to their routine mammography. The present study supports previous findings wherein a group of women with nearly seven times the average population risk of breast cancer was identified by the AI-based risk model.
Individualized Screening
Although about 6% of the women were high-risk, they are screened today in the same way as low-risk women. We think that a specially adapted screening could be more suitable for these women.
Mikael Eriksson, Study Leader and Postdoctoral Researcher, Department of Medical Epidemiology and Biostatistics, Karolinska Institutet
The aim of this study was to determine whether the approach, which has been evaluated in Sweden and the USA, is also effective in other mammography programs across Europe, rather than to investigate clinical application in and of itself.
Eriksson continues, “First you develop the model and test it in a slightly more limited population, and then you go on to demonstrate generalizability in other populations, after which you reach a point where you believe that the model works.”
The next stage of the project will involve a clinical trial in Europe where women will be examined, tested, and offered varying therapies based on the risk rating that the AI model assigns them. It was many years ago in the USA when this approach was clinically investigated.
Eriksson says, “We’re now looking at the possibility of introducing the model in Europe.”
The study is funded by the Swedish Research Council and the Swedish Breast Cancer Association. Mikael Eriksson owns a patent for an image-based risk model for breast cancer licensed to the US company iCAD, Nashua, NH.
Journal Reference:
Eriksson, M., et al. (2023). European validation of an image-derived AI-based short-term risk model for individualized breast cancer screening – a nested case-control study. The Lancet Regional Health - Europe, 100798–100798. doi.org/10.1016/j.lanepe.2023.100798.