Robotics in MRI-Guided Interventions : Topics in Magnetic Resonance Imaging

Secondary Logo

Journal Logo

Review Articles

Robotics in MRI-Guided Interventions

Hata, Nobuhiko PhD; Moreira, Pedro PhD; Fischer, Gregory PhD

Author Information
Topics in Magnetic Resonance Imaging: February 2018 - Volume 27 - Issue 1 - p 19-23
doi: 10.1097/RMR.0000000000000159
  • Free



Historically, medical robotics research used metal and electromagnetic components, precluding such robots to be used for in-bore procedures. The first modification to these procedures began when Masamune et al1 at the University of Tokyo used acrylic plastic for the frame and an ultrasonic motor as the actuator. The change in this material created an magnetic resonance imaging (MRI)-compatible motorized stereotactic frame for neurosurgery purposes. The research on MRI-guided robotics continued to evolve to include other stereotactic devices such as the 0.5T open-configuration scanner for prostate interventions,2 and the closed-bore scanner for breast interventions.3 According to Taylor and Stoianovici,4 these robots are classified as surgical computer-aided design and computer-aided manufacturing (CAD/CAM) robots. This is distinct from the class of robots referred to as “surgical assistant systems.” Unlike surgical assistant systems such as the da Vinci surgical system master-slave robot (Intuitive Surgical, Sunnyvale, CA), the primary goal of a CAD/CAM robot is the high precision guidance of surgical tools, such as biopsy needles where final placement of needle is still performed by the physician.

Since Masamune's work,1 many engineering prototypes of MRI-compatible robots have been presented for different MRI-guided interventions, such as for liver,5 kidney,6 prostate,7–13 brain,14,15 breast,3,16 and general percutaneous procedures.17

For instance, with growing interest in MRI-guided robotics and neurosurgical interventions, a group at the University of Calgary, Calgary, Canada developed the NeuroArm, a clinically realistic robot which is MRI-compatible using materials such as titanium, polyetheretherketone, and polymethylene.15 As attested in15 the robot is equipped with an MRI-compatible force/torque sensor for haptics feedback. So far, there has not been an MRI-compatibility test nor a clinical report using the robot in an MRI scanner's bore while patients are in the scanner. If this device has such a capability, it will be a significant first step toward enabling MRI-compatible robots to perform high-dexterity surgical maneuvers using a master-slave mechanism. Similar effort is under way by a group at Worcester Polytechnic Institute14 as is shown in Figure 1.14

MRI-compatible piezoelectrically actuated 7-DOF stereotactic neurosurgery robot intended for MR image-guided conformal ultrasonic interstitial thermal ablation of brain tumors.14 7-DOF, 7-degree-of-freedom.

A few researchers continue to explore the use of MRI-guided robots for breast intervention. Developed by researchers at the Karlsruhe Institute of Technology,33ROBITOM is an MRI-compatible robot for breast intervention. This robot is able to place a needle guide in 2 Cartesian directions using ultrasonic motors while the needle insertion is manually controlled. As of 2005, a second version of the ROBITOM robot1818a18 presented. More recently, researchers at the University of Twente presented a pneumatic-driven robot targeting breast interventions.16 Their robot is still in the development stage and patient studies have not been performed yet.

In our research team, we are proposing a physician-controlled guidance device and for orienting cryotherapy probes in image-guided cryotherapy of renal cancers19 as is seen in Figure 2.19 Keeping the device's minimal footprint in mind, we developed a body-mounted tool guidance device. The device comprises 2 ring-shaped motion stages stacked on each other, with the top ring stage holding the cryotherapy probe. The rotation stages can orient the cryoprobe with the skin incision point as the remote-center-of-motion. The phantom test indicated that the accuracy of probe placement is significantly better with the robotic instrument guide (4.05 mm) than without the robotic instrument guide (6.25 mm, P < 0.001), even in the presence of body motion. With the robotic instrument guide, the number of attempts and total procedure time to complete the cryotherapy probe placement reduced regardless of the presence of organ motions.

Body-mounted robot for MRI-guided cryoablation of abdominal organs.19


One of the most active areas of research in MRI-guided robotics is prostate intervention. Several MRI-compatible robots have been developed for MRI-guided prostate interventions in the past decade.7–13 An example of such research prototype is shown in Figure 3.7 Those robots vary from simple devices to fully autonomous systems, and were evaluated in clinical and preclinical tests.

MRI-guided 4-DOF parallel manipulator for in-bore prostate transperineal interventions.7 Courtesy of Dr Iulian Iordachita, Johns Hopkins University. DOF, degree-of-freedom.

The first clinical trial using a robot to access the prostate was developed by van den Bosch et al.20 The robot was used to transperineally implant gold seeds in 1 patient. Krieger et al21 also reported the first clinical trial of an MRI-compatible robot for transrectal prostate access, named APT II. That system is an improvement over their first prototype22 and combines manipulator mechanics, using a steerable needle channel and a hybrid tracking method. A comparison between robot-assisted and manual transrectal in-bore prostate biopsy was reported by Schouten et al.23 The clinical trials were performed with a pneumatic robot controlled in an open loop. The findings indicated that there was no significant reduction in the biopsy error or the procedure time. A group from Johns Hopkins University also report 2 cases of prostate interventions using their robot.24

One of the authors of this article (N.H.) reported clinical deployment of another robotic needle-guidance template for in-bore 3T transperineal MRI-guided prostate biopsy25 but in larger cohorts (Fig. 4).25 That study reports on the utility of a robotic needle-guidance template for in-bore 3T transperineal MRI-guided prostate biopsy. Specifically, it compares manual and robotic biopsies through a 2-arm mixed retrospective-prospective study. A 3T wide-bore MRI scanner (MAGNETOM Verio 3T; Siemens AG, Erlangen, Germany) was used to perform all biopsy procedures. The manual approach used a manual needle-guidance template, which comprises a grid of holes 1.3 mm in diameter spaced 5 mm apart. 3D Slicer (, a planning and navigation software, was used to plan which guide hole would be used for needle insertion and how deep to insert the needle for each core sample collection. The robotic approach used a motorized robotic needle-guidance template, which is able to set the needle insertion hole with a control resolution of 0.001 mm. During the robotic approach, the 3D Slicer software was modified to issue a motion control command to the robotic needle-guidance template to align the needle insertion hole to the transperineal needle placement trajectory. A confirmation image was taken using axial 2D multislice Turbo Spin Echo images after clinicians inserted the needle, as well.

Smart template device to place needle-guidance hole robotically, overcoming the coarse needle placement resolution in transperineal template with limited number of needle guide holes.25

Two ultrasonic motors with integrated optical encoders are used in the robot. The motor rotations are transmitted to vertical and horizontal prismatic motion of each cross bar by a lead screw-and-nut mechanism. A needle insertion position is created on the axial plane of the template by the cross point of the bars. The structural body of the motorized template is made of brass alloy screws and ultem resin. Timing belts, pulleys, and a miter-gear set compose the nonferrous transmission parts.

Tilak et al's study was conducted with 96 patients in 99 cases; 43 of these cases were performed using the robotic template and the remaining 56 cases were performed using the conventional needle guide template. A significant statistical difference existed in the log of accuracy when the best attempts of needle placement data were observed; the robotic approach 2.39 mm had statistically higher accuracy compared with the manual approach 3.71 mm by P < 0.027. Furthermore, the mean core procedure time for the manual group was 100.63 minutes, whereas the robotic group's core procedure time was only 90.82 minutes. In comparing the record of diagnosis per subject, no statistical difference existed between the manual and robotic groups; positive cancer was confirmed in 30 of 56 cases in the manual group (53.57%) and 25 of 43 cases in the robotic group (58.13%). In studying the diagnosis per core, however, cancer was detected in 174 of 541 cores (32.16%) from patients in the manual group, whereas 137 of 310 cores (44.19%) from the robotic group were cancer positive. This suggests that the robotic approach has a greater chance of providing a positive cancer core than the manual approach (P = 0.018).


Needle placement error can be modeled as a probability distribution, as elegantly described in references26 and 27 Assuming that increased biopsy accuracy is translated into the probability of the needle hitting the tumor, or the “hit rate,” we can use this as a measure for MRI-guided prostate interventions. Supposing that needle placement error follows the norm, the hit rate for a tumor with a volume of clinical significance (0.5 mL sphere with a radius of 4.92 mm) is estimated to improve from 63.9% to 98.6%, as the accuracy is improved from our previous clinical study 5.3925 to 2.0 mm. For next-generation biopsies, there are increasing demands for improved accuracy of prostate sampling (multiple foci within the gland). New research is underway to determine the combined ability of imaging and genomics to further characterize focal prostate cancer (and multiple foci within a single lesion), using a so-called radiogenomics approach. This approach shows early promise for examining possible intratumoral heterogeneity by biopsy and treating accordingly.28 In practice, the hit rate of aiming for a quadrant of lesion to deal with intratumoral heterogeneity will increase from 35.2% to 78.1%. If the accuracy of needle placement is improved from 5.3925 to 2.0 mm. For ablations, the clinical impact is measured by the ablation margin rather than the accuracy of needle placement. The needle placement accuracy is translated into an ablation margin to ensure the complete ablation of the tumor. The margin can be roughly estimated by using a simple model consisting of spherical tumor and ablation volume. If one aims to cover the tumor using a single probe with a placement accuracy of 5.39 mm, one would need to add 10.56 mm to the regular ablation margin (typically 5 mm) to achieve complete coverage at a 95% confidence level. The margin would be decreased to 3.92 mm if the needle placement accuracy is 2 mm. The scope of the current proposal is limited to assessing the improved accuracy using the suite of advanced tools. The efficacy of biopsy and ablation, namely, the hit rate and decreased ablation margin, will be assessed in follow-up clinical studies with a large cohort of patient enrollment.


Perhaps, the most advanced form of MRI-compatible robot, which has been developed and clinically validated, and commercialized, is INNOMOTION robot developed by Melzer et al,17 for percutaneous placement of needles in MRI scanner. This robot has a robotic arm attached to a C-shaped outer arm that tightly fits into the 60-cm bore of a standard MR scanner, guiding a biopsy needle. In Leipzig, Germany, Moche et al, performed 2 biopsies in the bone (femur and sacral bone), 3 other biopsies in different regions not subject to organ motion (lesser pelvis, iliac lymph node, lumbar spine abscess), and 1 in a region subject to respiratory motion (liver). The authors stated that a stereotactic device such as INNOMOTION is helping which will assist biopsies of lesions that are either hard to reach, difficult to align by mental mapping of the trajectory, or in regions with few anatomical reference points. The same principle have applies to most of the CAD/CAM robots in MRI-guided therapies. The system, under the commercial name INNOMOTION, was the first commercially available MRI-compatible robot for general percutaneous interventions, and received the Conformite Europeenne mark in 2005. However, the INNOMOTION robot is no longer in the market. Although the system was not specifically designed for prostate interventions, Zangos et al29 reported a cadaver study for transgluteal access to the prostate.


Cooperative Control

When cooperative robotic control becomes available, the clinician maintains ultimate control through a hands-on interface and the instrument moves only on the clinician's command; however, the motion will be autonomously directed to ensure that the desired path is followed to the intended target. We believe this is a sensible and safe next step in image-guided robotics to improve precision tool placement by taking advantage of novel robotic vision and automation technologies. Only limited investigation has been conducted on automated (actuated) needle insertion in surgical robotics, and even the published report on robotic needle insertion does not use imaging as feedback20,30 or was validated only in phantom studies.9 In our view as well as experts’ view in Nature Robotics article,31 the automation in medical robotics shall be introduced with significant attention to safety. The proposed technique leverages the capabilities of novel tele-operated semiautonomous needle placement in a clinically viable and safe way, as our approach integrates with a traditional workflow whereas the clinician maintains direct control of the intervention for safety. The other key innovation for safe and successful deployment of cooperative robot in this proposal is haptic sensing and display, which otherwise will be lost in teleoperation robots. Anecdotally, radiologists remark that haptic feedback is valuable during manual needle procedures, and there is some evidence that haptic display is useful in robot-assisted needle insertion.32 Also, in our initial patient studies where the robot acted as an automated guide for manual insertions, we noted that tactile sensation is critical (and should not be lost when going to actuated insertion). In other applications, haptic feedback in robot-assisted minimally invasive procedures has increased the physician's sense of presence and led to faster and more accurate task execution.33–36 In the past, frontiers of medical robotics have proposed an hands-on robotics for open surgeries. However, the solution to use hands-on cooperative robotics in real-time image-guided therapy is unprecedented, new, and in increasing demand in the dawn of precision medicine.

Full Autonomy

The autonomous medical robots are expected to be introduced to physicians with careful control by governmental regulatory bodies such as United States Food and Drug Administration (FDA). Most of the FDA-cleared medical robot today uses some form of artificial intelligence including image segmentation or motion control that automates part of physician's task to reduce their cognitive load. FDA however seems to be more comfortable to put physicians as ultimate decision maker by mandating them to review and approve the decision made by computers. The summary statements of FDA-cleared medical robots often say that devices are intended to “assist the surgeons” (MAKO Surgical Corporation, The Robotic Arm Interactive Orthopedic System, K093425) or “assist in the accurate control” of devices (Intuitive Surgical, da Vinci Endoscopic Instrument Control System, K040237), thus limiting the role of medical robots as assistant, not replacement to physicians. As role of robots evolves and autonomies are introduced, regulatory body will have to define how physicians are involved in more detail acknowledging the fact computers are in fact making part of medical decision.

In reality, the role of robot in decision-making process is substantial for instance in external radiation therapy.37 In external radiotherapies, a robotic eye can autonomously track the motion of moving cancer target and emit the radiation beam to the cancer target using 6 degree-of-freedom robotic arm (Accuray, Synchrony Reparatory Tracking System, K120233). The motion of the cancer is often induced by respiratory and cardiac motion or any involuntary movement by patients undergoing radiation therapy. By automatically recognizing the artificial fiducial marker placed in organs in real-time x-ray images, the robot can be controlled to aim the moving target to increase the effectiveness of the radiation treatment.38 This image-based detection and motion control is fully autonomous and does not involve physician intervention, thus considered a primitive form of robotic automation.

Open Data in Clinical Research

Industry researchers rarely share their results of safety and accuracy validation used for FDA market clearance for medical device. Regulatory body such as FDA is certainly involved to monitor these testings, but not required to share the data with public. For instance, the summary statement available for public about the aforementioned radiotherapy device simply states that a test was performed, without further detail. As autonomous robotics evolves, and is applied to medical devices and delivered to patient, public perception of robot should be carefully managed to avoid anxiety or confusion about autonomous technology. Therefore, transparency of safety assurance process in FDA, and sharing of the test data submitted for clearance may help public to understand the autonomous medical robotics better. Added value of this increased transparency of data submitted to FDA is accelerated development and increased safety of autonomous medical robots in development. Collectively, public will be benefited from open data policy of the government in particular FDA.


The robotics in MRI-guided intervention has been an active area of investigation. As the growing number of articles report engineering prototype of the robot, paucity of material is available to investigate the utility of the robots from clinical studies. There are, however, opportunities to further explore the potential benefit of robots through application of emerging technology such as autonomous control.


1. Masamune K, Kobayashi E, Masutani Y, et al. Development of an MRI-compatible needle insertion manipulator for stereotactic neurosurgery. J Image Guid Surg 1995; 1:242–248.
2. DiMaio SP, Pieper S, Chinzei K, et al. Robot-assisted needle placement in open MRI: system architecture, integration and validation. Comput Aided Surg 2007; 12:15–24.
3. Felden A, Vagner J, Hinz A, et al. ROBITOM-robot for biopsy and therapy of the mamma. Biomed Tech 2002; 47 (suppl 1 pt 1):2–5.
4. Taylor RH, Stoianovici D. Medical robotics in computer-integrated surgery. IEEE Trans Rob Autom 2003; 19:765–781.
5. Song SE, Tokuda J, Tuncali K, et al. Design evaluation of a double ring RCM mechanism for robotic needle guidance in MRI-guided liver interventions. In: 2013 IEEE/RSJ International Conference on Intelligent Robots and Systems. 2013:4078–4083.
6. Bricault I, Zemiti N, Jouniaux E, et al. Light puncture robot for CT and MRI interventions: designing a new robotic architecture to perform abdominal and thoracic punctures. IEEE Eng Med Biol Mag 2008; 27:42–50.
7. Eslami S, Shang W, Li G, et al. In-bore prostate transperineal interventions with an MRI-guided parallel manipulator: system development and preliminary evaluation. Int J Med Robot 2016; 12:199–213.
8. Su H, Shang W, Cole G, et al. Piezoelectrically actuated robotic system for MRI-guided prostate percutaneous therapy. IEEE ASME Trans Mechatron 2015; 20:1920–1932.
9. Moreira P, van de Steeg G, Krabben T, et al. The MIRIAM Robot: a novel robotic system for MR-guided needle insertion in the prostate. J Med Robot Res 2016; 1750006.
10. Abdelaziz S, Esteveny L, Renaud P, et al. Design considerations for a novel MRI compatible manipulator for prostate cryoablation. Int J Comput Assist Radiol Surg 2011; 6:811–819.
11. Muntener M, Patriciu A, Petrisor D, et al. Transperineal prostate intervention: robot for fully automated MR imaging—system description and proof of principle in a canine model. Radiology 2008; 247:543–549.
12. Goldenberg AA, Trachtenberg J, Yi Y, et al. Robot-assisted MRI-guided prostatic interventions. Robotica 2010; 28:215–234.
13. Tokuda J, Tuncali K, Iordachita I, et al. In-bore setup and software for 3T MRI-guided transperineal prostate biopsy. Phys Med Biol 2012; 57:5823–5840.
14. Li G, Su H, Cole GA, et al. Robotic system for MRI-guided stereotactic neurosurgery. IEEE Trans Biomed Eng 2015; 62:1077–1088.
15. Sutherland GR, Wolfsberger S, Lama S, et al. The evolution of neuroArm. Neurosurgery 2013; 72 (suppl 1):27–32.
16. Groenhuis V, Veltman J, Siepel FJ, et al. Stormram 3: a magnetic resonance imaging-compatible robotic system for breast biopsy. IEEE Robot Autom Mag 2017; 24:34–41.
17. Melzer A, Gutmann B, Remmele T, et al. INNOMOTION for percutaneous image-guided interventions: principles and evaluation of this MR- and CT-compatible robotic system. IEEE Eng Med Biol Mag 2008; 27:66–73.
18. Pfleiderer SO, Marx C, Vagner J, et al. Magnetic resonance-guided large-core breast biopsy inside a 1.5-T magnetic resonance scanner using an automatic system: in vitro experiments and preliminary clinical experience in four patients. Invest Radiol 2005; 40:458–463.
19. Hata N, Song S-E, Olubiyi O, et al. Body-mounted robotic instrument guide for image-guided cryotherapy of renal cancer. Med Phys 2016; 43:843–853.
20. van den Bosch MR, Moman MR, van Vulpen M, et al. MRI-guided robotic system for transperineal prostate interventions: proof of principle. Phys Med Biol 2010; 55:N133–N140.
21. Krieger A, Iordachita II, Guion P, et al. An MRI-compatible robotic system with hybrid tracking for MRI-guided prostate intervention. IEEE Trans Biomed Eng 2011; 58:3049–3060.
22. Krieger A, Susil RC, Ménard C, et al. Design of a novel MRI compatible manipulator for image guided prostate interventions. IEEE Trans Biomed Eng 2005; 52:306–313.
23. Schouten MG, Bomers JGR, Yakar D, et al. Evaluation of a robotic technique for transrectal MRI-guided prostate biopsies. Eur Radiol 2012; 22:476–483.
24. Ball MW, Ross AE, Ghabili K, et al. Safety and feasibility of direct magnetic resonance imaging-guided transperineal prostate biopsy using a novel magnetic resonance imaging-safe robotic device. Urology 2017; [Epub ahead of print].
25. Tilak G, Tuncali K, Song S-E, et al. 3T MR-guided in-bore transperineal prostate biopsy: a comparison of robotic and manual needle-guidance templates. J Magn Reson Imaging 2015; 42:63–71.
26. Kobayashi Y, Onishi A, Hoshi T, et al. Modeling of conditions where a puncture occurs during needle insertion considering probability distribution. In: 2008 IEEE/RSJ International Conference on Intelligent Robots and Systems. 2008:1433–1440.
27. Martin PR, Cool DW, Romagnoli C, et al. Optimizing MRI-targeted fusion prostate biopsy: the effect of systematic error and anisotropy on tumor sampling. In: Medical Imaging 2015: Image-Guided Procedures, Robotic Interventions, and Modeling. Vol 9415. International Society for Optics and Photonics; 2015: 94151J.
28. Vogelstein B, Papadopoulos N, Velculescu VE, et al. Cancer genome landscapes. Science 2013; 339:1546–1558.
29. Zangos S, Herzog C, Eichler K, et al. MR-compatible assistance system for punction in a high-field system: device and feasibility of transgluteal biopsies of the prostate gland. Eur Radiol 2007; 17:1118–1124.
30. Stoianovici D, Song D, Petrisor D, et al. “MRI Stealth” robot for prostate interventions. Minim Invasive Ther Allied Technol 2007; 16:241–248.
31. Sridhar AN, Hughes-Hallett A, Mayer EK, et al. Image-guided robotic interventions for prostate cancer. Nat Rev Urol 2013; 10:452–462.
32. Gerovich O, Marayong P, Okamura AM. The effect of visual and haptic feedback on computer-assisted needle insertion. Comput Aided Surg 2004; 9:243–249.
33. Pacchierotti C, Prattichizzo D, Kuchenbecker KJ. Cutaneous feedback of fingertip deformation and vibration for palpation in robotic surgery. IEEE Trans Biomed Eng 2016; 63:278–287.
34. Tholey G, Desai JP, Castellanos AE. Force feedback plays a significant role in minimally invasive surgery: results and analysis. Ann Surg 2005; 241:102–109.
35. Westebring-van der Putten EP, Goossens RHM, Jakimowicz JJ, et al. Haptics in minimally invasive surgery—a review. Minim Invasive Ther Allied Technol 2008; 17:3–16.
36. Meli L, Pacchierotti C, Prattichizzo D. Sensory subtraction in robot-assisted surgery: fingertip skin deformation feedback to ensure safety and improve transparency in bimanual haptic interaction. IEEE Trans Biomed Eng 2014; 61:1318–1327.
37. Buzurovic I. Robotic tumor tracking techniques in radiation therapy. Adv Robot Autom 2012; 1:
38. Ozhasoglu C, Saw CB, Chen H, et al. Synchrony—cyberknife respiratory compensation technology. Med Dosim 2008; 33:117–123.

autonomy; MRI compatibility; MRI-guided interventions; navigation; robotics

Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved.