AIM-HER2, an artificial intelligence (AI)-based tool integrated into the AISight® Image Management System, was evaluated for HER2 immunohistochemistry (IHC) scoring in a retrospective cohort of 101 equivocal (2+) breast cancer cases. Concordance between AIM-HER2 and pathologist assessments was observed in 87% of cases, with nearly half of discordant results adjudicated as acceptable by fluorescence in situ hybridization (FISH). These findings, presented by MedStar at USCAP 2024, support the feasibility of AI-assisted workflows for HER2 IHC scoring and suggest that AIM-HER2 may improve reproducibility in HER2-low and borderline cases, where inter-pathologist variability remains a significant challenge.
Conference
USCAP 2024