Real-time MRI-guided interventions using rolling-diaphragm hydrostatic actuators
Samantha Mikaiel1,2, James Simonelli3, David Lu1, Kyung Sung1,2, Tsu-Chin Tsao3, and Holden H Wu1,2

1Radiological Sciences, University of California, Los Angeles, Los Angeles, CA, United States, 2Biomedical Physics, University of California, Los Angeles, Los Angeles, CA, United States, 3Mechanical and Aerospace Engineering, University of California, Los Angeles, Los Angeles, CA, United States

Synopsis

In this work we investigate a new rolling-diaphragm-based hydrostatic actuator design to achieve smooth remote manipulation without fluid leakage for MR-compatible robotic systems. We show that the actuators exhibit negligible impact on MR image fidelity and SNR, the actuator provides a linear displacement response over the fluid lines, and we were able to use the master/slave actuator pair to insert and retract the needle in a phantom with no leakage and no noticeable friction issues. Our new rolling-diaphragm hydrostatic actuators can potentially enable physicians to remotely perform real-time MRI-guided interventions.

Introduction

Limited accessibility to the patient in bore has restricted the advancement of real-time MRI-guided interventions (1), MR-compatible robotic systems are a potential solution to overcome the limited accessibility. In particular, hydrostatic actuation is inherently MR-compatible and can be designed to be low-cost systems. In this work we investigate a new rolling-diaphragm-based hydrostatic actuator design to achieve smooth remote manipulation without fluid leakage. We evaluate its effects on MR images, characterize the linearity of motion of the actuator, as well as demonstrate feasibility of remotely manipulating instruments under real time MRI guidance.

Methods

[System Design] The standard piston-based hydrostatic actuators are susceptible to fluid leakage and friction issues, and we designed the new actuators (Fig. 1) using a rolling diaphragm(2) to fully seal the hydrostatic cylinder. The diaphragms and actuator casing were all created in house. The identical master/slave actuator pairs were then connected to closed fluid channels (filled with water) to transmit force and displacement into the bore.

[Imaging Assessment] All experiments were performed on a research-only intraoperative MRI scanner (3.0T Prisma, Siemens). The slave actuator was placed in the bore next to a sphere phantom (Fig. 2). Gradient echo (GRE) scans of the phantom with and without the actuator were acquired. The images were then inspected for artifacts and the SNR was calculated using the two-scan difference method(3). The process was repeated with a grid phantom, where dimensions of the imaged grid lines with and without the robotic system were measured to assess distortion.

[Linearity Assessment] A bench top test was performed by pushing the master actuator in gradual increments until the extreme followed by pulling it back, and measuring the displacement of the slave actuator using a pair of laser doppler displacement meters with a resolution of 0.635 micron. The slave actuator was then connected to an MR-compatible biopsy needle (Cook) and inserted into a gelatin phantom. A high-resolution (1x1x1mm3) 3D GRE scan was done with the needle in the fully retracted position, fully inserted position, and finally retracted again to measure the stroke of the actuator.

[Feasibility under real-time MRI] Based on visualization from a real-time GRE sequence at 2 frames/second, an operator controlled the master actuator from the end of the patient table to evaluate MR-guided insertion and retraction of the needle in a gelatin phantom.

Results

[Imaging Assessment] No images artifacts were observed with the actuator inside the bore (Fig. 3). The SNR difference with and without the actuator is on average only 1.2%. The grid phantom images showed no distortion.

[Linearity Assessment] The input of the master versus output of the slave displacement can be seen in Figure 4. Linearity is demonstrated for both cases, slave actuator moves 0.9756 mm to every 1 mm displacement applied by the master in the push direction. For the pull direction this ratio is 0.9467 mm to every 1 mm applied. The stroke was measured to be 17mm. Using the high resolution 3D MRI scan, the stroke was measured to be 16mm, where the 1-mm difference from the bench top measurement of 17mm is within the pixel resolution of the MRI scan.

[Feasibility under real-time MRI] Under real-time MRI guidance, the operator was able to easily and smoothly insert and retract the needle in the phantom from the end of the patient table repeatedly (Fig. 5)

Discussion and Conclusion

Our rolling-diaphragm hydrostatic actuators exhibit negligible impact on MR image fidelity and SNR. The actuator provides a linear displacement response over the lines in both the push and pull direction. This is important for the operator to be able to remotely and accurately manipulate the needle to target locations and avoid important structures. Finally, we were able to use the master/slave actuator pair to insert and retract the needle in a phantom with no leakage and no noticeable friction issues. Our new rolling-diaphragm hydrostatic actuators can potentially enable physicians to remotely perform real-time MRI-guided interventions.

Acknowledgements

NSF Graduate Research Fellowship Program

References

(1) Moche M, et al., JMRI 2010;31:964–74. (2)Whitney JP, et al., IEEE/RSJ Int. Conf. Intell. Robot. Syst. 2014. (3)Dietrich O,et al., JMRI 2007;26:375–85

Figures

Figure 1: Diagram of the master/slave actuator design (left) with a photo of the actuator with the needle (right). The brass rods on the master are used to insert and retract the rod, and therefore control the needle attached to the rod of the slave.

Figure 2: In-suite set up of the master/slave actuators, fluid lines, phantom, and real-time imaging display (a). (b) Close up of the experiment set up with the parts labeled.


Figure 3: 3D GRE image of the grid phantom (a) without and (b) with the actuator and the sphere phantom (c) without and (d) with the actuator.

Figure 4: Input displacement of the master versus output displacement of the slave of the diaphragm actuator in millimeters in both the push and pull directions.

Figure 5: Real-time image frames showing the actuator-driven needle (a) at initial position, (b) partially inserted, (c) fully inserted, (d) partially retracted, and (e) fully retracted.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
3588