 This paper proposes a novel approach to detecting regions of interest, ROIs, in whole slide images, WSI's. The approach uses a vision transformer model to identify ROIs in WSI's, which are then used for automating her 2-grading for breast cancer patients. The results show that the proposed method increases the diagnostic agreement to 99.3%, reduces the time to 15 seconds for automated her 2-grading, and improves the accuracy of her 2-grading compared to existing methods. This article was authored by M. D. Shakawat-Hossain, Gallag Mohamed Sharia, M. M. Mahbubbleside, and others.