Researchers at the University of Maine unveiled an AI tool, Context-Guided Segmentation Network (CGS-Net), designed to improve early breast cancer detection by analyzing both tumor tissue and surrounding contextual tissue.

The work reflects a broader shift from traditional slide-by-slide biopsy to digital imaging and AI-assisted screening, aimed at speeding diagnoses and shortening time to treatment.

CGS-Net could help alleviate global pathologist shortages by triaging patients and accelerating treatment in areas with limited pathology resources.

The American Cancer Society highlights a rising emphasis on early detection, contributing to a five-year survival rate improvement across cancers and underscoring the potential impact of faster AI-assisted diagnostics.

The team emphasizes that CGS-Net is designed to assist, not replace, pathologists, with the goal of reducing patient wait times after abnormal mammograms or positive biopsies.

Led by Jeremy Juybari and Josh Hamilton, the project argues that incorporating surrounding tissue improves prediction accuracy compared with existing AI detectors.

CGS-Net can process hundreds of digital tissue images simultaneously to help triage patients more quickly after a positive biopsy by differentiating cancerous from normal tissue.

By integrating contextual information from tissue around a suspected tumor, CGS-Net may enhance diagnostic performance and aid in cancer confirmation, though it is not definitive cancer confirmation based on tissue type alone.

Breast cancer remains a major U.S. health concern as the second most common cancer among women, with about one in eight women at risk in her lifetime, and early detection is linked to higher survival rates.