Bibliometric and knowledge-map analysis of research on robot-assisted vascular interventional surgery (2015-2025)

Vascular interventional procedures are minimally invasive but remain highly dependent on operator expertise and are associated with occupational radiation exposure and technically demanding catheter and guidewire manipulation. By integrating precision motion control, force sensing, haptic feedback, image-guided navigation, and teleoperation, robot-assisted vascular interventional surgery (RVIS) has emerged as an interdisciplinary field spanning robotics, biomedical engineering, and interventional medicine. However, its global research landscape, intellectual base, research hotspots, and thematic evolution remain insufficiently characterized. This study therefore systematically mapped the knowledge structure and research trends of RVIS from 2015 to 2025. The Web of Science Core Collection (WoSCC) and Scopus were searched for relevant English-language articles and reviews published between 2015 and 2025. After screening according to uniform eligibility criteria, 373 WoSCC records and 493 Scopus records were included. VOSviewer, CiteSpace, and Bibliometrix/Biblioshiny were used to analyze annual publication trends, country and institutional collaboration, authors, journals, co-citation relationships, highly cited publications, citation bursts, and keyword networks separately for the two databases. After cross-database deduplication, an English-abstract corpus was constructed from 541 unique publications, of which 524 met the eligibility requirements for latent Dirichlet allocation (LDA) modeling. The model fitted to the merged corpus served as the primary integrative analysis, while database-source sensitivity analyses were conducted in the WoSCC-only and Scopus-only corpora using the same five-topic configuration. Multiple testing in the temporal topic-trend analyses was controlled using the Benjamini-Hochberg procedure. Annual publication output in both databases showed an overall upward, albeit fluctuating, trend and remained relatively high during the later years of the study period. China and the United States were the two most prominent contributors to the field. In the corresponding-author country analysis, China ranked first and the United States second in WoSCC, whereas China and the United States ranked jointly first in Scopus. Analyses of core journals, authors, and co-citation networks indicated that the intellectual base of the field was jointly shaped by robotics and control technologies, biomedical engineering, computer-assisted surgery, and cardiovascular and neurovascular interventions. Highly cited publications, citation bursts, and keyword networks primarily focused on magnetically actuated guidewires and soft robots, endovascular catheter control, intelligent navigation, force and haptic feedback, and robot-assisted percutaneous coronary intervention. LDA identified five topics, of which remote image-guided platforms had the highest overall topic prevalence. After correction for multiple testing, the topic of magnetically actuated soft robots and microrobots showed a significant upward trend in the merged corpus. The single-database sensitivity analyses demonstrated moderate reproducibility of the principal topics, with the annual trend for magnetically actuated soft robots and microrobots showing a consistent direction across all three corpora. Research activity in RVIS increased overall between 2015 and 2025, accompanied by the emergence of an interconnected knowledge structure spanning engineering technologies and clinical interventional medicine. Catheter and guidewire manipulation, remote intervention, intelligent navigation, force sensing, and haptic feedback represent major research directions. The relative research attention devoted to magnetically actuated soft robots and microrobots increased consistently across all three LDA corpora. These findings reflect changes in publication output and research attention and should not be interpreted as direct evidence of technological maturity, clinical adoption, or patient benefit.