Significant findings that reshape the landscape of early breast cancer detection have emerged from a study published in the journal "Radiology" by the Radiological Society of North America. The research, featuring Greek scientists Pantelis Gialias and Apostolia Tsirikoglou, demonstrates that artificial intelligence (AI) systems can identify early signs of the disease up to six years before a formal diagnosis, and in some instances, as early as a decade before.

The Precision of Data

The retrospective study utilized Swedish data, analyzing nearly 89,000 mammographies from 31,394 women between 2008 and 2019. Researchers evaluated three commercially available AI systems, which demonstrated a high discrimination capability (90%) between true positive and negative results.

  • In 20% of cases, AI recognized signs six years prior to diagnosis.
  • In 25% of cases, signs were visible four years earlier.
  • In approximately 15% of women, systems detected cancer up to ten years before diagnosis.

Beyond the Human Eye

According to Dr. Gialias, a radiologist and director at the Breast Center of Mediterranean Hospital, AI identifies "underlying changes where the human eye cannot say with certainty that something suspicious exists." These may include minor architectural distortions or a gradual increase in breast density—subtle markers that often elude even experienced radiologists.

"Essentially, it is an additional tool, a weapon we have for the early diagnosis of breast cancer. Naturally, all decisions must be made carefully with the necessary studies, and human oversight must always be present," Dr. Gialias emphasizes.

Efficiency and the Future

The research also highlights the economic and operational dimensions of the technology. Previous studies by Dr. Gialias indicated that AI could reduce radiologist workloads by up to 34%, while serving as a cost-saving strategy compared to the traditional method of double-reading by two doctors. The research team’s next step involves studying women with silicone implants, a group that presents additional imaging challenges.