The Evolving Landscape of Deep Learning in Breast Cancer Imaging: A Bibliometric Study of Segmentation, Detection,...

IntroductionDeep learning has rapidly reshaped breast cancer imaging, but the evolution of segmentation, detection, and diagnostic research remains insufficiently characterized. This bibliometric review mapped global trends, collaboration patterns, thematic evolution, and emerging frontiers from 2006 to 2025.MethodsThis bibliometric study retrieved publications on deep learning in breast cancer imaging from the Web of Science Core Collection and Scopus. English-language articles and reviews published between 2006 and 2025 were included. After deduplication, Bibliometrix, VOSviewer, and CiteSpace analyzed publication trends, country contributions, collaboration patterns, keyword co-occurrence, temporal topic evolution, and citation bursts.ResultsA total of 3,568 publications were included. Annual output remained limited before 2016 but increased markedly thereafter, with especially rapid growth after 2020, indicating the transition of this field from an exploratory stage to accelerated development. China ranked first in corresponding-author publications, whereas the USA and the United Kingdom showed stronger citation impact, reflecting differences between publication scale and academic influence. Keyword analysis showed that the field was primarily structured around deep learning, breast cancer, mammography, segmentation, detection, and computer-aided diagnosis. Temporal analyses further indicated a shift from early computer-aided diagnosis frameworks and conventional neural-network approaches toward more advanced and clinically relevant themes, including explainable artificial intelligence, nomogram, self-attention, transformers, neoadjuvant therapy, and axillary lymph node metastasis.ConclusionDeep learning in breast cancer imaging has evolved into a rapidly expanding and increasingly sophisticated field centered on segmentation, detection, and clinically relevant diagnostic research. Future progress will depend on improved interpretability, robust validation, and stronger clinical integration.