Interactive multistage robotic positioner developed for intraoperative MRI-guided stereotactic neurosurgery

A joint multidisciplinary team of researchers from the Chinese University of Hong Kong (CUHK), the University of Hong Kong (HKU), University College London, and Johns Hopkins University has developed an interactive, multistage robotic positioner for intraoperative MRI-guided stereotactic neurosurgery aimed at treating patients with neurological diseases, such as brain tumours and Parkinson’s disease.

In conventional stereotactic neurosurgery, guiding cannula or needle insertion with static preoperative MRI images may introduce errors in stereotactic frame set-up and image registration, and lead to the occurrence of brain shift (brain deformation during neurosurgical operation). [Neuroimage Clin 2017;17:794-803; Surg Neurol2001;56:357-364] “By reducing errors in conventional stereotaxis, our robotic positioner can enhance precision of instrument positioning and ensure [desired] surgical outcomes,” highlighted Dr Danny Chan of the Department of Surgery, CUHK.

The robotic positioner is a lightweight (203 g), compact (diameter, 97 mm; height, 81 mm), polymer-based structure that provides a sufficient workspace (±35°) for instrument targeting. Designed for skull-mounted usage, the system allows up to two robots to be fitted into most standard imaging head coils. [Adv Sci (Weinh)2024;11:e2305495]

In the robotic system, instrument orientation adjustment follows a two-stage, semi-automated approach. In the first stage, the surgeon orients the robot instrument guide based on preoperative images. The second stage involves automated soft robotic fine adjustment with finite element analysis–based optimization. During the process, orientation is secured by granular jamming and tendon-driven braking.

In laboratory testing, the robotic system demonstrated precision (<0.2° orientation error), responsiveness (1.4 Hz) and high resolution (0.058°) in soft instrument positioning. The integration of orientation locking provided sufficient transmission stiffness (4.07 N/mm) for instrument advancement.

Through validation with a skull model across target depths of 50–90 mm, the robotic system achieved an overall average accuracy of 0.73 mm for radial error (ie, the distance between the target and insertion trajectory). No observable image artifacts were detected in MRI images.

In evaluation of the system using a 1.5 T MRI scanner, the average accuracy for radial error was 1.7 mm with an agar gel–fabricated brain phantom and 2.2 mm with a defrosted cadaver sample. These results suggested that the system met the clinical design target of <3 mm.

After instrument insertion and target verification, real-time MRI can be used to monitor surgical procedures. In the robotic system, custom-made wireless omnidirectional tracking can facilitate robot registration under MRI. Moreover, zero electromagnetic interference generation allows the use of intraoperative MRI guidance during robot actuation and for evaluating interventional processes.

“Preparations are being made for clinical tests of the robotic positioner,” added Chan. “Additionally, the first intraoperative MRI system in Hong Kong is set to start providing service in the third quarter of 2024, with plans to install two or three additional systems within the next 5 years.”

Robotic system implementation within an MRI head coil. Adapted from Adv Sci (Weinh) 2024;11:e2305495