AI evaluation of MRI scan helps rule out cancer in women with dense breasts
While mammograms have reduced deaths by detecting breast cancers when they’re small and easier to treat, it’s less effective for women with dense breasts.
However, a new study finds that supplemental MRI screening can make a difference for these women, who are more likely to develop breast cancer. And new technology is being used to speed the process.
Artificial intelligence can quickly and accurately sift through MRIs to rule out breast cancer in the majority who don’t have it — freeing up radiologists to work on the more complex cases, Dutch researchers report.
In the Dense Tissue and Early Breast Neoplasm Screening, or DENSE, trial, investigators trained artificial intelligence technology to distinguish between breasts with and without lesions.
“The DENSE trial showed that additional MRI screening for women with extremely dense breasts was beneficial,” said lead author Erik Verburg, of University Medical Center Utrecht in the Netherlands. “On the other hand, the DENSE trial confirmed that the vast majority of screened women do not have any suspicious findings on MRI.”
Mammography is less sensitive in women with extremely dense breasts than in women with fattier tissue. Women with extremely dense breasts also have as much as six times the risk of developing breast cancer compared to women with fatty breasts. Their risk is twice that of the average woman.
The study analyzed MRIs of nearly 9,200 extremely dense breasts. Of those, more than 8,300 had no growths and 838 had at least one. Of those, 77 were cancerous
The model flagged 91% of the MRIs with lesions for a radiologist’s review. It dismissed about 40% of the lesion-free MRIs without missing any cancers, according to the study.
The findings were published this week in the journal Radiology.
“We showed that it is possible to safely use artificial intelligence to dismiss breast screening MRIs without missing any malignant disease,” Verburg said in a journal news release. “The results were better than expected. Forty percent is a good start. However, we have still 60% to improve.”
Verburg said this AI-based system has the potential to significantly reduce radiologists’ workload.
“The approach can first be used to assist radiologists to reduce overall reading time,” Verburg said. “Consequently, more time could become available to focus on the really complex breast MRI examinations.”