Researchers at Google alongside Northwestern University and three other medical institutions of Britain are working on an AI model. This AI model is capable of detecting breast cancer in human mammograms.
The study was published in the journal Nature, on the New Year’s Eve. The study states that the AI model was capable of detecting breast cancer better than radiologists. The data was also proven internationally, which helps this one to differentiate itself from the previous ones.
Only an expert radiologist with years of training the eye is able to detect the masses in a mammogram. Even such trained eyes fail to detect cancer at times. According to the American Cancer Society, 20 percent of breast cancers go undetected in women. Therefore, alternate sources that help correct diagnosis will be a tool of the future. As such, AI cannot replace the human eye at least on the field of detecting cancers. Instead, they will help in forming a second opinion in the art of corrective diagnosis.
The AI tool was a collaboration between researchers from Google Health, DeepMind, along with Northwestern University experts. In addition, radiologists from the UK Imperial Center Cancer Research and Royal Surrey County Hospital collaborated with the study. Initially, the researchers fed the algorithm with 91,000 cases of breast cancer of women from the US and UK. The clinical data was only fed and rest all were kept anonymous.
The AI was then tested on news sets breast scans in women from both countries. The results obtained from the algorithm were mindblowing. In the US where women are screened every year the AI detection had 6% fewer false alarming reports and 9% lesser false positive reports. While in the UK where screening is done every 3 years, AI detection had 1% fewer alarming reports and3% fewer false positives.
The results are a breakthrough as they are superior to the standards of clinical study in both countries. In addition, the AI was challenged against six radiologists to detect 500 breast cancer scans, in which the AI won. This may be a keyhole in the art of diagnosis of future diseases with AI. One should also note that places AI failed to detect the cases where doctors detected it and vice Versa. Although such technologies make a gateway in diagnosis, the future of such tools’ usage isolation is a question mark. In addition, AI cannot replace the human interaction between clinicians and patients, such a complication is only answerable when both work synergistically.
Read our summary of the most popular stories of last week, in on our newsletter, What Just Happened!
Apply to write for us if you love writing and want to see your articles featured on Apple News, Google News, and more. Subscribe to our newsletter, What Just Happened, where we dive deep into the hottest topics from the week!