Google's AI system can beat doctors in breast cancer detection


A study that tested the accuracy of the system, which was developed through a collaboration between the tech giant and cancer researchers, was published Wednesday in the scientific journal Nature.

The program was trained to detect cancer using tens of thousands of mammograms of women in the United Kingdom and the United States, and early research shows that it can produce more accurate detection than human radiologists.

According to the study, the use of AI technology resulted in less false positives, where test results suggest that cancer is present when it is not, and false negatives, where an existing cancer is not detected.

Compared to human experts, the program reduced false positives by 5.7% for EE subjects. UU. And 1.2% for subjects in the United Kingdom. Reduced false negatives by 9.4% for US subjects. UU. And at 2.7% for subjects in the United Kingdom.

The AI ​​system was more accurate despite having less information to work than human experts, such as patient histories and previous mammograms.

Professor Ara Darzi, one of the authors of the article and director of the Imperial Cancer Research Center in the United Kingdom, said he did not expect to see such an impressive result of the AI ​​system. "This is one of those transformative discoveries that he has in his hands, which could alter the way we perform the detection in terms of improving accuracy and productivity," Darzi told CNN Business.

Breast cancer is the second leading cause of cancer death in women, although the results improve considerably if the disease is detected and treated early, say the study authors.

However, despite large-scale breast cancer screening programs in developed countries, screening mammograms do not detect one in five breast cancers, according to the American Cancer Society.

"The performance of even the best doctors leaves room for improvement," the study authors wrote in Nature. "AI can be specially prepared to help with this challenge."

The technology could also be used to address the shortage of radiologists, they say.

A 2018 report from the Royal College of Radiologists found that 75% of the directors of the UK radiology department feel that there are not enough clinical radiologists to provide a safe and effective level of patient care.

The report forecasts that about 2,000 additional radiologists will be required to close the projected deficit in 2023.

The car's camera system could help keep drivers awake at the wheel

In the United Kingdom, where two radiologists interpret each mammogram, the AI ​​system could be used to replace the second reader, Darzi said. "I don't think this replaces humans at this stage," he said.

Breast cancer screening and diagnosis are found alongside other Google research that demonstrates how AI can help doctors more accurately predict lung cancer and prevent serious eye diseases.

The study published Wednesday was part of a collaboration between Google Health, Cancer Research UK Imperial Center, Northwestern University and Royal Surrey County Hospital.