AI helps radiologists detect breast cancer in real-world tests

According to the largest study of its kind, artificial intelligence models can actually help detect cancer and reduce doctors’ workload. Radiologists who chose to use AI were able to identify one additional case out of 1000 cases of breast cancer.

Alexander Katalinik and his colleagues at the University of Lübeck, Germany, worked with nearly 200 certified radiologists to test an AI trained to identify signs of breast cancer from mammograms. Radiologists examined 461,818 women at 12 breast cancer screening sites in Germany between July 2021 and February 2023, and each one could choose whether to use AI or not. This resulted in 260,739 patients examined by an AI and a radiologist, with the remaining 201,079 patients examined by a radiologist alone.

Those who decided to use AI successfully detected breast cancer at a rate of 6.7 cases per 1,000 scans – 17.6 percent higher than the 5.7 cases per 1,000 scans among those who chose not to use AI. Is. Similarly, when women underwent biopsy after a suspected diagnosis of cancer, among those diagnosed with AI, 64.5 percent of women were likely to have a biopsy where cancer cells were found, compared with 59.2 percent of women diagnosed with AI. Was not used.

The scale at which AI improved breast cancer detection was “extremely positive and exceeded our expectations,” Catalinic said in a statement. “We can now demonstrate that AI significantly improves cancer detection rates in breast cancer screening.”

“The goal was to show non-inferiority,” says Stephen Bunk of Vara, an AI company involved in the study. “If we can show that AI is not inferior to radiologists, that’s an interesting scenario for saving some workload. “We were surprised that we were able to show excellence.”

Over-reliance on AI in medicine has some people concerned as it could lead to some symptoms of a condition being missed, or a two-track system of treatment where those who can pay are afforded the luxury of human contact. Can be done. There was some evidence that radiologists spent less time examining scans that the AI ​​had already suggested were “normal” – meaning cancer was unlikely to be present – ​​reviewing them for an average of 16 seconds. While the AI ​​could do on them in less than 30 seconds. Don’t categorize. But these latest findings have been welcomed by those specializing in the safe deployment of AI in medicine.

“The study provides further evidence for the benefits of AI in breast screening and should be another wake-up call for policymakers to accelerate AI adoption,” says. ben glocker Imperial College London. “The results confirm what we are seeing again and again: With the right integration strategy, the use of AI is both safe and effective.”

He welcomes the way the study allowed radiologists to make their own decisions about using AI, and would like to see more trials of AI in a similar manner. “We can’t easily assess this in the laboratory or through simulations and instead need to learn from real-world experience,” says Glocker. “The technology is ready; Now we need to follow the policies.”