AI predicts interval cancer risk based on negative mammograms
Artificial intelligence-based tools have the potential to predict whether a woman will develop interval breast cancer based on their negative imaging, opening doors for earlier treatment and improved outcomes.
That’s according to new research published in RSNA’s flagship journal Radiology. The study results suggest a novel algorithm can predict up to 43% of interval cancers before they actually develop. This sort of risk assessment could give providers an opportunity to send patients for supplemental imaging that could offer more detailed visualization of areas AI flags as suspicious.
“Personalized breast cancer screening depends on accurately assessing an individual’s risk of developing breast cancer within a specific timeframe,” said Fiona J. Gilbert, MB ChB, a professor in the department of radiology at the University of Cambridge, United Kingdom. “We can use supplemental imaging and adjust screening frequency based on a woman’s breast density and likelihood of developing breast cancer within a short timeframe.”
The ability to predict interval cancers is intriguing for providers, as these cases are often more aggressive and have worse prognoses. In areas where annual screening is not the norm, such as the U.K., where triennial screening is recommended, AI assessments of interval cancer risk could be especially beneficial.
Gilbert and colleagues recently retrospectively evaluated a deep learning algorithm’s (Mirai) potential for predicting three-year interval cancer risk. They deployed the tool in a U.K. screening cohort of over 134,000 patients across two sites. The algorithm assigned risk scores to each exam, with higher percentiles representing those at greatest risk of developing cancer within three years after their initial baseline screening. Its predictions were compared alongside patients’ electronic health records to determine accuracy.
Researchers determined the algorithm accurately predicted interval cancers in 43% of the women who it assigned its highest risk score, 20%. For those assigned 1%, 5% and 10%, the algorithm predicted 3.6%, 14.5% and 21.6% of interval cancers. It was most accurate when predicting cancer onset within one year of the completion of baseline imaging, the group noted.
“Our results suggest that further workup of mammograms within the top 20% of scores could yield 42.4% of interval cancers, meaning that Mirai could be used to identify women for supplemental imaging or a shortened screening interval, instead of or in addition to breast density,” suggested lead researcher Joshua W. D. Rothwell, an MBBS/PhD student at the University of Cambridge.
The team conducted an additional analysis on the effect of breast density on the algorithm’s accuracy. That assessment revealed that accuracy decreased in instances of extremely dense breasts. However, its predictions still outperformed conventional risk assessments, indicating it has utility for flagging women who would benefit from supplemental imaging or shortened screening intervals.
Read more about the team’s findings here.
In addition to her background in journalism, Hannah also has patient-facing experience in clinical settings, having spent more than 12 years working as a registered rad tech. She began covering the medical imaging industry for Innovate Healthcare in 2021.