It seems Volpara aren't the only people out there using AI to screen mammograms. Though not reported widely, I shave read of many research projects and companies exploring the use of "AI" in radiology. The majority of radiological images are static, black and white, pixels, 1s and 0s, so it shouldn't really be a surprise. Volpara might be the most commercially advanced currently but I suspect they will have to spend mightily to stay ahead of the competition.
Disc - don't hold, sold 3 years ago or so.
From the Economist
When a new book on artificial intelligence—co-written by Henry Kissinger—hit the shelves in November, one of its main examples was research from 2020 that used AI to discover a new antibiotic. It works better than existing drugs because it attacks the bacteria in a novel way, to which they have not built up resistance. Behind the life-saving algorithm are Regina Barzilay, a computer scientist at the Massachusetts Institute of Technology (MIT), and a team of researchers. It is just one of several areas where she uses AI to improve medicine.
Born in Moldova, Dr Barzilay was educated in Israel and America and began her academic life in computational linguistics. She developed algorithms to classify news stories by theme and to summarise them. But in 2014, in her mid-40s, she was diagnosed with breast cancer. After undergoing chemotherapy and lumpectomies at Massachusetts General Hospital, she was aware that just nearby, in her academic department, technology existed that could have helped her and could save the lives of millions of women around the world. From that moment, she changed her research to focus on AI in health care. A MacArthur “genius” grant followed in 2017.
Her work has made contributions to the field beyond antibiotics. In recent months Dr Barzilay, along with Adam Yala, a doctoral student at MIT, has been working to introduce an AI system called Mirai, which can predict whether a patient is at risk of breast cancer within as much as five years. Having scanned thousands of mammograms, the AI is able to uncover subtle, intricate patterns that elude human eyes and cognition. “If this technology is used in a uniform way,” she says, “we can identify early who are high-risk patients and intervene. The earlier the cancer is detected, the easier it is to treat it and the outcomes are much better,” she says.
The system she built might have benefited her: after her first mammogram she was informed that she was healthy. Three years later she was diagnosed with cancer. When she fed that early mammogram into Mirai, the AI indicated she had been at high risk. Too late to warn her, but Mirai may alert many others.
