Accurate histopathological diagnosis is critical for effective breast cancer treatment planning, prognosis, and personalized patient care. Histopathological image analysis, leveraging computational pathology and artificial intelligence (AI), has demonstrated significant potential in the automated classification and personalized breast cancer treatment. This paper critically reviews five prominent publicly available breast cancer histology datasets—TCGA-BRCA, BreaKHis, ICIAR BACH, BRACS, and BreCaHAD. We comprehensively discuss dataset characteristics, including patient cohorts, histopathological classifications, magnification levels, annotation methodologies, and genomic and clinical information integration. Strengths and limitations are systematically highlighted, emphasizing dataset size, annotation granularity, class diversity, and multi-modal integration capabilities. Comparative analyses provide clarity for researchers selecting datasets aligned with their research goals. Lastly, we outline existing gaps and propose future directions, advocating for improved dataset annotation, reduced demographic biases, and enhanced integration of histological data with clinical and molecular markers, thus supporting more effective personalized breast cancer treatment through computational methods.