AI-Assisted and Robotic Ultrasound for Vein and Vascular Mapping: A Scoping Review of Clinical Applications, Automation and Haemodialysis Access

This scoping review will map the current evidence on artificial intelligence-assisted, automated, autonomous and robotic ultrasound technologies relevant to pre-operative haemodialysis-access vessel mapping. Pre-operative vascular mapping for haemodialysis access requires more than recognition of a vessel within an individual ultrasound image. A clinically useful examination may require appropriate image acquisition, identification of arteries and veins, continuous tracking of named vessels, assessment of vessel calibre and depth, recognition of branches, tributaries and perforators, dynamic and Doppler assessment where appropriate, evaluation of image quality and examination completeness, and reconstruction of the findings into a clinically actionable vascular map. Artificial intelligence, computer vision and robotic ultrasound systems have demonstrated a growing range of individual capabilities, including vessel segmentation, automated measurements, vessel tracking, robotic probe manipulation and three-dimensional vascular reconstruction. However, these component technologies should be distinguished from an end-to-end system capable of performing a complete clinical vessel-mapping examination. The primary objective of this review is therefore to determine which components of a clinically complete pre-operative haemodialysis-access ultrasound vessel-mapping examination have been demonstrated using artificial intelligence, computer vision, robotic ultrasound or autonomous ultrasound acquisition, and which required clinical competencies remain insufficiently validated. The review will distinguish direct evidence from haemodialysis-access and AVF/endoAVF applications from enabling evidence derived from closely related vascular-ultrasound technologies. Particular attention will be given to acquisition, probe control, anatomical recognition, continuous vessel tracking, quantitative measurement, vascular topology, Doppler and dynamic assessment, image-quality assurance, uncertainty and failure detection, examination completeness, three-dimensional reconstruction, map generation and the continuing requirement for expert sonographer input. The review will follow current JBI methodological guidance for scoping reviews and will be reported in accordance with PRISMA-ScR. The anticipated output is an evidence map of current technological capabilities and a proposed evidence-informed, sonographer-led competency framework describing the capabilities that would require validation before an AI-assisted or autonomous system could be evaluated against expert sonographer-performed pre-operative vessel mapping.