Someday, computers could help doctors diagnose diseases much faster than they can today. Researchers from Beth Israel Deaconess Medical Center (BIDMC) and Harvard Medical School (HMS) have developed a way to train artificial intelligence to read and interpret pathology images. Andrew Beck from BIDMC explains that their "method is based on deep learning," which is commonly used to train AI to recognize speech, images and objects. They recently got the chance to prove just how effective their technique is during a competition at the annual International Symposium of Biomedical Imaging, where the AI was tasked to look for breast cancer in images of lymph nodes.
The team started training their machine by feeding it hundreds of slides marked to indicate which parts have cancerous cells and which have normal ones. They then identified which types of slides it was having the most trouble with and fed it more difficult samples. Using that method, the AI improved enough to be accurate 92 percent of the time and to win in two separate categories during the contest. It's still no match for human pathologists who are accurate 96 percent of the time, but it's clearly shown great promise.
Beck said what's truly exciting is that when they combined pathologists' analysis with their creation's, the results showed 99.5 percent accuracy. He added: "Our results in the ISBI competition show that what the computer is doing is genuinely intelligent and that the combination of human and computer interpretations will result in more precise and more clinically valuable diagnoses to guide treatment decisions." If you want to read more about this breast cancer-detecting AI, the team published a paper detailing their experience.