AI Helps Detect Breast Cancer Faster?

Good news for all. 

Early results from a Swedish clinical trial show that AI can be used to detect breast cancer. And it does so 20 percent better than traditional screening of mammograms by radiologists.

Published in the Lancet Oncology, using randomized, controlled, population-based trial, with women aged 40–80 years at four screening sites in Sweden, it involved more than 80,000 subjects.

While half of the participants had their mammograms inspected by two radiologists, the other half has them screened  by AI and a radiographer. Whenever the AI generated the highest risk score, two radiologists assessed the mammograms manually.

The key finding is that using AI-supported analysis of mammograms, with one or two radiologists, was as effective as involving two radiologists on their own. Thanks to AI, 20 percent more cancers were detected. 

In practical terms this means that radiologists can spend significantly a lot less time on screening mammograms and helping patients.

Although these findings are positive, lead author Kristina Lång of Lund University warns against jumping to conclusions just yet. She says that these results “are not enough on their own to confirm that AI is ready to be implemented in mammography screening.” Accordingly, more trials will be held before, and if, the program is fully rolled out.

Stephen Duffy, professor of cancer screening at Queen Mary University of London, also commented on the study. While he emphasised that it is good that the radiologists would have time to help more patients, AI might also end up over-detecting harmless lesions.

The study’s results play into the notion that AI could significantly boost research in critical areas like healthcare, as suggested by Microsoft’s vice chair and President Brad Smith.

Previously, Gagarin News reported that AI is treading on EU translators’ heels.