Artificial intelligence can be used in breast cancer screenings to improve diagnosis and ease the workload of radiologists, according to the results of a clinical trial in Sweden. But researchers say preliminary results of the study, completed in August, show the technology is not yet ready to replace monitoring by radiologists.
At Lund University in Sweden, about one million women a year undergo mammographic screening, say its leaders.
The X-ray images of the breast are then usually analyzed by two radiologists to see if there are signs of cancer, a time-consuming method called “double reading”. A radiologist can analyze an average of about 50 examinations per hour.
But can technology help ease this workload?
Researchers in southern Sweden are evaluating the effect of AI-supported examinations. This is the first trial of its kind comparing the two diagnostic methods.
“First we analyze the examinations without using Artificial Intelligence, using our advantages like the possibility to see the whole image. Then we use the device with Artificial Intelligence and then we can analyze the signs in the image”, says Kristina Lang, leader of the study and professor of radiology at Lund University.
In the clinical trial, which involved more than 80,000 women, researchers used Artificial Intelligence to identify “high-risk” breast cancer screenings, which were then subjected to traditional double readings by radiologists.
The results of the trial revealed that AI-powered screening of mammograms resulted in a similar cancer detection rate to the standard double-reading method by radiologists.
This almost halved the workload of doctors, reading data on the screen was reduced by about 44 percent.
Ms Lang says this level of time saving can be vital, particularly in understaffed areas.
“In low- and middle-income countries there is a huge shortage of doctors. So these data-analysis algorithms can make a really big difference in regions where there are no experts,” she says.
In Copenhagen, in neighboring Denmark, AI has been used in daily mammography examinations since November 2021. The results of the study there were so impressive that Danish doctors in May 2022 allowed the use of AI for the first reading of mammograms of “low-risk” patients, about 70 percent of all examinations, which are then analyzed by an experienced radiologist.
“I think it’s exciting because we have a shortage of doctors across the country, especially mammography radiologists. Now, we have a valuable tool to make their job easier,” says Dr. Ilse Vejborg, who participated in the decision to use Artificial Intelligence as a method to detect breast cancer in Copenhagen.
The Swedish researchers say their next step is to learn which types of cancer were detected with and without the support of Artificial Intelligence. 100 thousand women have already registered for this clinical trial.
Further studies are needed before the use of Artificial Intelligence becomes commonplace, but for patients like 82-year-old Elisabeth Colin, the technology raises hopes that cancer symptoms can be detected at earlier stages.
“I hope that you will be able to dictate changes much earlier than now, so that people can be helped in a different way,” she says.
According to the World Health Organization, 2.3 million women were diagnosed with breast cancer in 2020 and 685 thousand died from this disease worldwide.