HONOLULU, Hi. — Artificial intelligence does a better job of identifying the hidden signs of breast cancer, scientists in Hawaii say. Their study finds a sophisticated tool, a type of AI called deep learning, analyzes breast scans to identify women with early signs of cancer whose tumors are hard to find by conventional methods.
Doctors recommend that all women over the age of 40 book themselves for a mammogram at least once a year to screen for breast cancer. These images help physicians detect the cancer early, significantly increasing a patient’s chances of survival.
They also help to determine how likely a woman is to develop cancer further down the line, based on her breast density. While dense breasts on a mammogram signifies a higher risk of cancer, there are many other, unknown factors, which doctors can’t account for. Now, researchers at the University of Hawaii have taught a computer to look for these hidden signals.
“Conventional methods of breast cancer risk assessment using clinical risk factors haven’t been that effective,” says lead author Professor John Shepherd in a statement to SWNS. “We thought that there was more in the image than just breast density that would be useful for assessing risk.”
Making the yearly screening process easier?
The team analyzed more than 25,000 breast scans from 6,369 women. They found over 1,600 women had developed breast cancer that was successfully picked up during their screening. However, another 351 developed interval invasive breast cancer, meaning it was not detected during the regular screening process.
Scientists trained a “deep learning model” to find hidden details or signals in the scans which could indicate a higher risk of breast cancer. The tool was more effective than traditional methods at predicting the cancers which were picked up during regular screenings. On the other hand, the technology was less effective when trying to calculate the risk of cancers developing between screenings.
“The results showed that the extra signal we’re getting with AI provides a better risk estimate for screening-detected cancer,” Professor Shepherd tells SWNS. “It helped us accomplish our goal of classifying women into low risk or high risk of screening-detected breast cancer.”
Armed with the algorithm, women could be sorted into groups depending on their level of risk instead of having to come back a year later for another screening.
“This would allow us to use a woman’s individual risk to determine how frequently she should be monitored,” Shepherd says. “Lower-risk women might not need to be monitored with mammography as often as those with a high risk of breast cancer.”
Smarter technology dealing to smarter cancer decisions
“Women in the high-risk deep learning group who also have dense breasts and are at a higher risk for interval cancers may benefit most from a monitoring strategy that includes supplemental imaging that retains sensitivity in dense breasts such as MRI, ultrasound and molecular imaging,” the study authors explains. “Interval cancers usually have more aggressive tumor biology and are typically discovered at an advanced stage.”
Next, the researchers are hoping to replicate the study in Native Hawaiian and Pacific Islander women, groups which have been underrepresented in breast cancer studies until now. They are also hoping to study different grades of breast cancer, from the least to most aggressive forms.
“By ranking mammograms in terms of the probability of seeing cancer in the image, AI is going to be a powerful second reading tool to help categorize mammograms,” Prof. Shepherd concludes.
The findings appear in the journal Radiology.
South West News Service writer Tom Campbell contributed to this report.