A new peer-reviewed study has explored whether AI tools that are good at spotting breast cancer in mammograms can also be good at figuring out a woman’s risk of developing breast cancer in the future.
Researchers looked at mammogram images of 3,386 women, some of whom were diagnosed with breast cancer and others who were not, to test this idea. They used these images to see if the AI could have predicted the cancer risk before it was actually detected.
The study used a special kind of AI model, named Mirai, along with three other AI tools originally designed to detect cancer, to see if they could also assess the risk of getting cancer.
They checked how well these tools worked by comparing their predictions against actual cases of cancer detected during screening and also looked at how these predictions matched up with cancer risks assessed from earlier mammograms taken three years before a diagnosis.
The findings showed that the Mirai AI model was the best at both detecting cancer and assessing cancer risk. There was also a pattern noticed where AI tools that were better at finding bigger cancers were also better at predicting cancer risk. This suggests that improving how well AI tools can spot smaller cancers might make them better at predicting cancer risk too.
The researchers concluded that AI tools used for detecting breast cancer could potentially be tweaked to help assess cancer risk, which could be very useful for personalizing screening programs. If a woman’s risk of developing cancer could be accurately predicted, she might receive recommendations for more frequent screenings or other preventive measures.
However, the study also noted some limitations. For instance, they couldn’t explain exactly why the AI models made certain predictions, and the study only looked at mammograms from certain types of machines. Despite these limitations, the study’s findings offer a promising step toward using AI not just for detecting breast cancer but also for predicting it, which could lead to more personalized and effective screening strategies.
Mirai is an advanced AI (Artificial Intelligence) tool specifically designed to assess the risk of breast cancer using mammogram images. In more technical terms, Mirai uses a deep-learning algorithm, which is a sophisticated type of machine learning. This algorithm analyzes the visual details in mammograms, which are X-ray images of the breast, to predict how likely it is that a woman will develop breast cancer within the next five years.
Deep-learning algorithms like the one used in Mirai are capable of identifying complex patterns in images that might not be visible to the human eye. Mirai works by examining each mammogram in detail, processing the images through four separate modules within the algorithm. Each module focuses on different aspects of the images, and then Mirai combines the insights from these modules to estimate a woman’s annual risk of developing breast cancer for the next five years.
One of the innovative aspects of Mirai is its ability to process just the image data to make its predictions, though it can also incorporate other risk factor data if available. This makes Mirai particularly versatile and powerful because it can provide risk assessments based on the mammograms alone, offering a potentially important tool for doctors and patients in managing and monitoring breast cancer risk.
The technical prowess of Mirai lies in its deep-learning foundation, which allows it to learn from vast amounts of data and improve its predictions over time. By training on large datasets of mammogram images, Mirai has learned to detect subtle indications of future cancer risk that may not be apparent in traditional screening methods. This capability could revolutionize how breast cancer screening and risk assessment are approached, providing a more personalized and proactive strategy for early detection and prevention.