An Artificial Intelligence (AI) program seems capable of reducing the workload of radiologists in breast cancer detection, according to the first data from a study published on Wednesday, although it is too early to conclude on its effectiveness.
Conducted in Sweden and published in 'The Lancet Oncology', this study also concludes that there is no risk of radiologists using artificial intelligence (AI) software to better target their analyses.
The researchers divided about 80,000 women into two groups of similar size.. All of them SE underwent a mammogram, but the first group was screened conventionally, that is, with the eyes of two independent radiologists, while the data from the second were examined first by an AI and then by a single radiologist.
The AI-assisted group had no worse results and even detected a slightly higher number of cancers.
On the other hand, the rate of “false positives” – cases in which the first test erroneously detects cancer – was similar.
Using this technology could cut the workload for these doctors in half because the procedure involving artificial intelligence only needs 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 between the ages of 50 and 74, in line with European recommendations.
“The big potential for AI right now is that it could allow radiologists to be less stressed by excessive amounts of reading” of analysis, said Kristina Lang, a radiologist at Sweden's Lund University and lead author of the study.
But Lang believes that these promising interim results are not “sufficient on their own to confirm that AI is ready to be implemented in mammography detection,” he said in a statement.
Despite the study, it is too early to conclude the interest of AI in this field and it will take several years to know if it has been as effective as human double opinion.
In this sense, the researchers plan to compare within two years the rate of cancers that have escaped detection but will have been diagnosed.
The study data do not shed light on the risk of “overdiagnosis,” that is, the detection of lesions that would not have developed into dangerous cancers without treatment.
The issue of “overdiagnosis” is the focus of criticism against the widespread testing policy, although research is increasingly confirming that it does serve to reduce mortality from breast cancer.
The risk of overdiagnosis “should encourage caution regarding the interpretation of the results,” cautioned oncologist Nereo Segnan, who was not involved in the study, in a comment to The Lancet Oncology, while acknowledging that it is promising.
Stephen Duffy, a professor of cancer screening at Queen Mary University of London who was not involved in the study, notes that AI may have overdiagnosed certain forms of early breast cancer, called ductal carcinoma in situ.
However, he praised this “high-quality study” and said that reducing the workload of radiologists is “an issue of considerable importance” in many breast cancer screening programs.