Breast Screenings Using AI Boost Cancer Detection

Share on twitter
Share on linkedin
Share on facebook
Share on reddit
Share on pinterest

Preliminary findings from a landmark Swedish trial indicate that artificial intelligence (AI) is aiding radiologists in more efficiently detecting breast cancer cases. Scientists discovered that AI-supported breast screening detected 20% more cancers than the traditional method of two radiologists double-reading mammograms. Derived from a study of over 80,000 women, these results were recently published in The Lancet Oncology journal.

Mammogram plays a crucial role in early cancer detection, making treatment more effective. Interestingly, introducing AI to the screening process didn’t increase the occurrence of false positives or incorrect identifications of routine mammograms as abnormal. Furthermore, this approach significantly reduced the screen-reading workload for radiologists by 44.3%.

Dr. Kristina Lång, the study’s lead author from Lund University in Sweden, expressed her surprise with the results. She stated that the findings exceeded the research team’s expectations. She emphasized the potential of a screening method that is accurate and also efficient.

While there’s global apprehension about AI replacing human roles in many sectors, Dr. Lång clarified that AI isn’t aiming to replace radiologists in this context. She highlighted radiologists’ pivotal role, especially in ensuring a minimal false positive rate.

While AI is adept at flagging potentially suspicious findings, human expertise is essential to confirm if these flagged cases are genuine instances of cancer.

The European Commission’s guidelines advocate for two radiologists’ double reading of mammograms. However, there’s an overwhelming workload for radiologists due to a shortage in many countries. This is where AI can play a supportive role, highlighting potential concerns while radiologists make the final calls.

In the study, over 80,000 women were divided into two groups. Half underwent AI-assisted breast screening. The other half had their mammograms double-read by two radiologists without AI assistance between April 2021 and July 2022.

In the AI-aided group, the system gauged the cancer risk on a scale of 1 to 10. Two radiologists analyzed mammograms with the highest risk score (10), whereas those scoring below ten were examined by just one.

Preliminary data reveals that 244 cancers were detected in the AI group, with 861 women recalled due to suspicious findings. Conversely, the group without AI saw 203 cancers, with 817 women recalled.

However, experts Nereo Segnan and Antonio Ponti, who were not involved in the study, emphasize caution. They raised potential concerns about overdiagnosis with AI-supported screening.

Another limitation the study’s authors acknowledged was the confinement of the analysis to one facility using a specific mammography device and AI system.

As Dr. Lång pointed out, the ongoing study’s objective is to assess cancers, focusing on their type and stage comprehensively. This in-depth analysis will aid in determining whether AI can genuinely revolutionize the early detection of breast cancer.

0 replies on “Breast Screenings Using AI Boost Cancer Detection”

Related Post