Three‐dimensional ultrasound‐guided robotic needle placement: an experimental evaluation

Abstract

Clinical use of image-guided needle placement robots has lagged behind laboratory-demonstrated robotic capability. Bridging this gap requires reliable and easy-to-use robotic systems. Our system for image-guided needle placement requires only simple, low-cost components and minimal, entirely off-line calibration. It rapidly aligns needles to planned entry paths using 3D ultrasound (US) reconstructed from freehand 2D scans. We compare system accuracy against clinical standard manual needle placement. The US-guided robotic system is significantly more accurate than single manual insertions. When several manual withdrawals and reinsertions are allowed, accuracy becomes equivalent. In ex vivo experiments, robotic repeatability was 1.56 mm, compared to 3.19 and 4.63 mm for two sets of manual insertions. In an in vivo experiment with heartbeat and respiratory effects, robotic system accuracy was 5.5 mm. A 3D US-guided robot can eliminate error bias and reduce invasiveness (the number of insertions required) compared to manual needle insertion. Remaining future challenges include target motion compensation.