Development of a shoulder-mounted robot for MRI-guided needle placement: phantom study

Document Type

Journal Article

Publication Date



International Journal of Computer Assisted Radiology and Surgery








MRI-compatible robot; Patient-mounted; Percutaneous procedures


© 2018, CARS. Purpose: This paper presents new quantitative data on a signal-to-noise ratio (SNR) study, distortion study, and targeting accuracy phantom study for our patient-mounted robot (called Arthrobot). Arthrobot was developed as an MRI-guided needle placement device for diagnostic and interventional procedures such as arthrography. Methods: We present the robot design and inverse kinematics. Quantitative assessment results for SNR and distortion study are also reported. A respiratory motion study was conducted to evaluate the shoulder mounting method. A phantom study was conducted to investigate end-to-end targeting accuracy. Combined error considering targeting accuracy, respiratory motion, and structure deformation is also reported. Results: The SNR study showed that the SNR changes only 2% when the unpowered robot was placed on top of a standard water phantom. The distortion study showed that the maximum distortion from the ground truth was 2.57%. The average error associated with respiratory motion was 1.32 mm with standard deviation of 1.38 mm. Results of gel phantom targeting studies indicate average needle placement error of 1.64 mm, with a standard deviation of 0.90 mm. Conclusions: Noise and distortion of the MR images were not significant, and image quality in the presence of the robot was satisfactory for MRI-guided targeting. Combined average total error, adding mounting stability errors and structure deformation errors to targeting error, is estimated to be 3.4 mm with a standard deviation of 1.65 mm. In clinical practice, needle placement accuracy under 5 mm is considered sufficient for successful joint injection during shoulder arthrography. Therefore, for the intended clinical procedure, these results indicate that Arthrobot has sufficient positioning accuracy.

This document is currently not available here.