Artificial Intelligence can detect breast cancer – 8/2/2023 – Science

An Artificial Intelligence (AI) program appears to be able to reduce the workload of radiologists in detecting breast cancer, according to the first results of a study published this Wednesday (2), although it is too early to conclude on its effectiveness. .

Made in Sweden and published in The Lancet Oncologythe survey also concludes that there are no risks in using artificial intelligence to guide radiologists’ analyses.

The researchers separated about 80,000 women into two similarly sized groups. All underwent a mammogram. The first group was followed up in the conventional way, that is, through the eyes of two independent radiologists, while the data from the second group were analyzed first by an AI and then by a single radiologist.

The analysis done by AI did not perform worse and even detected a greater number of tumors in its group. On the other hand, the rate of “false positives”—cases in which the first test falsely detects cancer—was similar.

The use of this technology could halve the workload of these physicians because the procedure involving AI needs only one radiologist.

promising results

These results are promising because screening tests are one of the main ways to fight breast cancer. In countries like France, tests are very common among women aged 50 to 74, following European recommendations.

“The great potential of AI right now is that it could reduce the stress on radiologists from overanalyzing” scans, said Kristina Lang, a radiologist at Lund University in Sweden and lead author of the study.

However, Lang believes these promising preliminary results are not “sufficient by themselves to confirm that AI is ready to be implemented in screening mammograms,” he said in a statement.

Despite the study, it is too early to conclude the performance of AI in this field and it will take several years to know if it is as effective as human double opinion.

In this sense, the researchers intend to compare, within two years, the rate of cancers diagnosed that were not detected in the analyses.

The study data shed light on the risk of “overdiagnosis”, that is, the detection of tumors that would not have become dangerous without treatment.

The issue of “overdiagnosis” focuses criticism against the widespread testing policy, even though research confirms its role in reducing mortality from breast cancer.

The risk of overdiagnosis “should awaken caution regarding the interpretation of the results,” warned oncologist Nereo Segnan, who was not involved in the study, in a statement to The Lancet Oncologyalthough he recognized that the study is promising.

Stephen Duffy, professor of cancer detection at Queen Mary University of London, who was not involved in the study, suggests that the AI ​​may have overdiagnosed certain forms of early breast cancer, called carcinoma. ductal in situ.

However, he praised the “high quality study” and said reducing radiologists’ workload is “an issue of considerable importance” in many breast cancer screening programs.