This AI detects cancer in almost 100 % of cases, it does better than doctors

A new AI model has a nearly 100% detection rate when it comes to spotting signs of cancer from microscope images. Its results are impressive.

Artificial intelligence is particularly gifted when it comes to making a medical diagnosis. Capable of analyzing millions of images and signs in a few seconds, it acts with a precision that can never be matched by the human brain. This is why we are arriving at AI capable of telling if you are sick simply by looking at your tongue. Or here, of affirming with almost 100% certainty if a patient has endometrial cancer (the lining of the uterus) by examining images of its cells.

“The proposed ECgMLP model outperforms existing methods by achieving an accuracy of 99.26%,” summarizes Dr. Asif Karim, co-author of the study presenting the AI. For comparison, based on the same images, doctors show a detection rate of around 80%. The gain is therefore very significant. Especially since the model is easily adaptable to other types of cancer. The teams have already done this and the results are just as impressive.

Almost 100% success in cancer detection, this AI impresses with its efficiency

“We evaluated the model on several histopathological image datasets. It diagnosed colorectal cancer with an accuracy of 98.57%, breast cancer with an accuracy of 98.20% and oral cavity cancer with an accuracy of 97.34%,” explains Professor Niusha Shafiabady, another co-author of the study. In all cases, it is better than what a health professional is capable of doing. There is no question of replacing them, however. The tool must be considered as a diagnostic aid.

Read also – This AI sees the future, it detects if you are at risk of a heart attack in 10 years

Faster and more affordable than current systems, AI will above all allow for earlier patient care. If cancer is detected early, the chances of effective treatment increase, as do the patient's chances of survival. It remains to be hoped that the democratization of this very promising technology will not take too long to arrive.

Source: Charles Darwin University