BOSTON (WHDH) - Cancer survivor Dr. Regina Barzilay is a cancer survivor who is using her experience to help others.
The professor at the Massachusetts Institute of Technology helped develop a way for artificial intelligence to analyze mammograms, with an algorithm that can detect small changes that might otherwise go unnoticed, which means possibly being able to predict developing cancers up to five years earlier.
“Once we finished developing the system we looked back at my own mammograms and it’s quite clear that my cancer was seen by machines at least two years before my diagnosis,” Barzilay said.
Now, 10 years cancer-free, Barzilay said her cancer battle led her to study ways to use AI predictive technology.
Her new AI too, called MIRAI, allows women to get a higher resolution breast cancer screening.
It also analyzes mammograms to assess future risks, especially in aggressive cancers that affect younger women.
The tool can also indicate whether a person is at a higher risk for developing cancer even before it develops.
“The doctor actually looks and says, ‘okay, there is some mass here, we need to maybe biopsy or investigate’,’ Barzilay said. “Here, we’re looking at this kind of woman that’s currently considered not having the disease, they send home, but the model can now know how, actually, these women are high-risk.”
Brazilay is among three researchers who just received the WebMD Health Heroes award for their medical breakthroughs using AI.
She said it is critical for patients to stay informed on their health, so they can catch any problems in time.
“We do have a right to know our risk and then we, together with our healthcare providers, need to manage them,” she said.
The team said MIRAI identified nearly two times more future breast cancer diagnoses than the current clinical standard.
Brazilay said she hopes AI predictive tool scan also be applied to other cancers in the future.
(Copyright (c) 2024 Sunbeam Television. All Rights Reserved. This material may not be published, broadcast, rewritten, or redistributed.)