AI detects breast cancer more often and faster than radiologists: ‘Good collaboration can save lives’ | Nina’s Instagram

Breast cancer is the most common form of cancer in women. One in every 9 women is infected with the disease, and approximately 2,280 citizens die from its consequences every year. Preventive examination is crucial for the success of treatment. This has led to the Population Breast Cancer Survey, where women aged 50 to 69 are encouraged to get screened every two years.

Breast examinations (mammograms) performed by radiologists take a lot of time and money. For this reason, the results of Swedish research in the field of artificial intelligence, published in the scientific journal The Lancet Oncology, are considered promising. AI is already being used in radiology, but not yet in decision-making. The new AI software now under investigation was created by Dutch company ScreenPoint, and has been trained to detect breast cancer from mammograms.

More than 80,000 women participated in this pioneering research. Half of their scans were examined by two radiologists. The other half was evaluated by artificial intelligence software. Results? AI detects breast cancer more often than doctors, detects fewer false positives and cuts the workload in half. Breast cancer is detected again for every thousand women. This may not seem like a lot, but every year in Belgium alone, this may include hundreds of women who may have been misdiagnosed.

“The combination of detecting breast cancer more often and faster without additional false positive reports is particularly interesting,” says Iggy van de Poel, an expert in the field of big data in healthcare. “This does not change the fact that the work of radiologists has been and will continue to be valuable. The greatest promise lies in good cooperation between radiologist and AI. Non-practicing radiologist and AI expert Sietske Rozie can confirm this. “The focus should be on the doctor’s awareness of taking Medical decisions using feedback from artificial intelligence.”

Due to rising costs as well as the increasing number of radiological tests that radiologists have to review, it is not always possible to have enough radiologists to check the scans. Because of this “shortage” of radiologists and an aging population, the profession is currently experiencing a high workload. Artificial intelligence offers a solution to this. Not only does artificial intelligence detect breast cancer more often than radiologists, but computers also do it almost twice as fast. New technology can reduce the significant workload of radiologists.

When looking at x-rays, the same question is always asked: Is there breast cancer, yes or no? Artificial intelligence can answer this question without the need for a radiologist. “Everyone gets tired at some point, including the radiologist, while AI can check mammograms continuously,” Siitsky-Rosi explains.

“AI is not yet a replacement for humans, but in the near future there will be almost no radiologist working without the help of AI,” says Eric Tetero, radiology coordinator at the Population Breast Cancer Survey. The goal now is to find out how artificial intelligence can support the doctor and make him better and smarter. “But the radiologist’s work shouldn’t be completely taken over by the AI. You then run the risk that the radiologist won’t end up understanding the AI, whereas the computer should always be able to be examined by the radiologist,” Rosie adds.

The new technology for breast cancer screening has not yet been used in our hospitals. “Analytical research shows that unfamiliarity with capabilities and uncertainty about added value, among other things, play an important role,” says Rosi. In short, there is a lack of awareness, knowledge and familiarity with technology.

The trust of both medical professionals and patients in AI is crucial to the success of the collaboration. Because do women wait for computerized mammograms?

A Dutch study recently showed that most women between the ages of 16 and 75 would prefer not to have their mammograms evaluated by AI alone. Of the 922 women surveyed, 77.8 percent agreed with the need for a human examination. “Understandable,” Rosie says. “We all know that computers sometimes suffer from malfunctions.” The results of this study give hope. “It is important that all employees in hospitals are introduced to the phenomenon of AI in an accessible way. I am working hard on this now. All healthcare professionals should receive training on how their minds work and how AI can influence their decisions, both positively and negatively on Both. If doctors and staff have confidence in working with AI, patients will too.

This article previously appeared in Libelle NL.

After her two sisters, Nelly Rayman will soon undergo a preventive mastectomy

The number of people surviving breast cancer is increasing, but can the disease be completely eliminated? “Metastatic breast cancer is a completely different story” (+)

Sigrid (63 years old) with breast cancer: “The oncologist asked me how much alcohol I drink, I said a glass every day” (+)