The Artificial intelligence It has been postulated in a short time as a technology with enormous revolutionary potential in many areas of our lives, and medicine was not going to be less. In many different ways, scientists and doctors are applying it to improve treatment and diagnosis of, for example, many types of cancer.
Spots that turn into cancer
Thus, a group of researchers from the Lauder Breast Cancer Center of the Memorial Sloan Kettering Cancer Center (United States) is using an algorithm designed by scientists from the Massachusetts Institute of Technology (also in the United States) to predict the cases of breast cancer up to four years before for them to develop.
As explained by Dr. Larry Norton, the institution’s medical director, in an interview with the American media CNN, the key is to detect a series of spots that, in the future, can develop and become breast cancer.
These spots, he pointed out, are not in themselves a sign of the initial appearance of the cancer itself, but should serve as a signal to the radiologists to be forewarned and begin to carry out a closer follow-up or even make decisions with the patient herself.
Reduce errors to zero
The algorithm was developed using a large bank of mammography images from healthy patients who either later developed breast cancer or never did at all. In this way, and as artificial intelligences usually operate, the system look for common patterns in all those who fell ill (and that they do not appear in those that did not).
In Norton’s opinion, such a technology could help reduce the number of diagnostic errors, which is currently considerable in breast cancer, to zero: up to one in eight. It must also be remembered that, as with most tumors, early detection of breast cancer It greatly improves the patient’s prognosis.
References
Poppy Harlow. Images show AI detecting breast cancer 4 years before it developed. CNN Health (2023). Viewed online at https://edition.cnn.com/videos/health/2023/03/07/artificial-intelligence-breast-cancer-detection-mammogram-cnntm-vpx.cnn on 03/23/2023.