AI predicts breast cancer progression with pathologist-level accuracy

Two international studies published in The Lancet Oncology, led by the Peter MacCallum Cancer Centre, have demonstrated that artificial intelligence can predict the progression of breast cancer with accuracy comparable to expert pathologists. The research involved the analysis of tissue samples from over 5,600 patients.

The studies focused on evaluating the density and spatial distribution of tumor-infiltrating lymphocytes (TILs) within tumor tissues. High levels of these immune cells indicate a stronger defense mechanism and a more positive response to treatment. By analyzing data from the CATALINA and APHINITY clinical trials, researchers found that AI models performed with high precision in both ‘triple-negative’ and ‘HER2-positive’ aggressive breast cancer subtypes.

Beyond simple cell counting, the AI technology mapped the spatial distribution of immune cells, identifying “hot spots” that are difficult for the human eye to quantify under a microscope. This capability helps predict how well patients might benefit from specific targeted therapies, particularly in cases involving lymph node involvement.