Artificial intelligence more accurate than the doctors diagnose breast cancer on mammography results, BBC reports citing a study published in the journal Nature.
An international team of scientists, in particular, Google Health and Imperial College London have developed a computer model and taught her how to diagnose cancer on the basis of x-ray images almost 29,000 women.
It turned out that the algorithm has read mammograms better than six separate radiologists .
Diagnostics of artificial intelligence proved to be as accurate as the joint work of two doctors.
Also, in contrast to humans, artificial intelligence never gets tired. Experts say that it can be used to improve the diagnosis of cancer.
How good is artificial intelligence?
In modern British health system x-rays every woman analyzed by two radiologist. In rare cases, when they can’t agree, evaluates images of the third doctor.
In the study, artificial intelligence, analysed the anonymous pictures, which it was impossible to identify the women. Unlike human experts who have access to the history of patients before artificial intelligence was only mammograms.
The results showed that the insights of artificial intelligence was as accurate as the diagnostic scheme with the participation of two doctors.
At the same time, artificial intelligence was diagnosed with cancer better than a single doctor. Compared to a single radiologist, the AI was 1.2% fewer false positive results (when mammogram to see cancer, which actually is not).
Also 2.7% less false negative results when the cancer is not diagnosed.
“Our team is really proud of the results of these studies, suggesting that we are on the way to the development of a tool that helps to diagnose breast cancer with greater accuracy,” said Dominic king from Google Health.
In order to become a radiologist, which can decipher the mammogram, you need more than 10 years of preparation. Reading radiographs is a vital but very time-consuming work.
Replace if the artificial intelligence people?
No. After all, people have developed a model of artificial intelligence and taught her how to read images.
This was a pilot study and now the AI system directly in the clinics have not used responsible for the diagnosis, even if the AI will be at least one radiologist.
But the AI could largely eliminate the need for double reading of mammograms by two doctors, thus reducing their workload, researchers say.
Professor Professor Darzi Ara, co-author of the report and Director of the Imperial centre for cancer research UK (CRUK), told the BBC: “This has gone far beyond my expectations. This will have a significant impact on improving the quality of reporting and free up radiologists more time to do more important things.”
Women aged 50 to 70 years in the UK should be screened for breast cancer every three years.
The use of AI may eventually speed up diagnosis, since the computer algorithm can analyze the image for a few seconds.
Sarah Hiom, Director of research and early diagnosis of cancer in CRUK, told the BBC: “This promising early study which suggests that in the future it will be possible to make the diagnosis more accurate and efficient, which will mean less waiting and anxiety for patients, as well as the best results.”
Helen Edwards from Surrey was diagnosed with breast cancer at the age of 44 years before she won the right for a free screening. She needed surgery, chemotherapy and radiation therapy, but she overcame cancer, and since then it’s been more than a decade.
She represented the patients in the panel discussion CRUK, which will have to decide whether to allow Google Health using anonymising data about breast cancer.
“At first I was a bit concerned with what aim Google may use the data, but they lack any identifiers. In the long term this can only benefit women, she told the BBC. Machines of artificial intelligence do not get tired… they can work around the clock, while man cannot do so, because combining them together is a great idea.”