5.1 Introduction

In 1992, the ROBODOC system (Integrated Surgical Systems, Sacramento, CA) was the first machine to partake in a human-robotic surgery. Since then, many surgical subspecialties including urology, general surgery, and gynecology have worked to continue to expand robotic use. Robotics has been increasingly utilized in spine surgery in an effort to mitigate some of the surgical risk.

The advantages of technology and robotics to the surgeon has been consistently demonstrated in the form of fatigue reduction, increased surgical precision, and even decreased postoperative pain. ¹ Shared-control robots have been used to assist in varying spine procedures including pedicle screw placement, tumor resections, and vertebroplasty.

Kosmopoulos and Shizas performed a meta-analysis of 16,717 pedicle screws placed in the lumbar spine in vivo and found that 86.7% were accurately positioned. Specifically, navigated pedicle screw placement improved screw accuracy by a mean of 5% when compared to traditional freehand techniques. ² Automation of surgery via robotics has provided promising results and as more studies emerge, its benefits have become more evident.

5.2 Robotic Surgery for Transpedicular Fixation

Transpedicular fixation has been the primary, and most widely practiced and studied use, for shared control robots. Robotics and navigation can help mitigate some of the challenges of pedicle screw placement, including pedicle fracture, pedicle screw misplacement, and radiation in the setting of fluoroscopic assistance.

One of the original descriptions of shared control robotics for spine application was by Sautot et al in Grenoble, France, in 1992, ³ who adapted an industrial robot to assist in transpedicular fixation. Preoperative planning employed computed tomographic (CT) scans to generate three-dimensional (3D) images of segmentation of a particular vertebra which served as a guide for pedicle screw trajectory. Intraoperatively, two X-ray devices formatted a 2D projection. The robot carrying a laser optical guide was then employed to superimpose a laser beam in the preoperatively planned surgical trajectory. The authors subsequently demonstrated a drilling experiment on plastic vertebrae, with reported submillimeter accuracy. ³

From 2002 to 2012, eight out of 18 novel spine surgical robots developed had a particular focus on transpedicular screw insertion. ⁴ In 2003, a team of Israeli investigators developed the MiniAture Robot for Surgical Procedures (MARS), ^{5 ,​ 6} which ultimately became the SpineAssist/Renaissance system, currently commercialized by Mazor Robotics (Caesarea, Israel). The revolutionary innovation of MARS/SpineAssist was a significant reduction in size and weight which allowed for direct attachment to the patient. These miniature robotic systems allowed for a drastic simplification of the process where the surgical robot interpreted the spatial reference points directly based on the patient’s bony landmarks.

5.3 SpineAssist Mazor Robotics

The SpineAssist robot/Mazor system (MAZOR Robotics Inc, Orlando, FL) has been the most widely studied of all shared control robots utilized in spine surgery. Conceptually utilizing three separate outrigger arms with an accommodating drill guide sleeve, the robotic software, in sync with a computer-assisted navigation, predicts which arm allows for the most accurate pathway for pedicle instrumentation determined by the implant and relative location of the SpineAssist robot to the predetermined entry point and screw trajectory. The robot may be attached and positioned directly onto a spinous process or it may be affixed to a frame triangulated by percutaneously placed guide wires (e.g., one Kirschner wire [K-wire] attached to the spinous process and two Steinmann pins in the posterior superior iliac spines) in the setting of minimally invasive spine (MIS) procedures. ⁷

In the preoperative period, the SpineAssist software creates a virtual spinal map by procuring and registering CT images of the operative spinal levels. Subsequently, templating of the desired screw entry point, trajectory, screw diameter, and length can be performed. The templating process may be done preoperatively or intraoperatively and is derived from the 3D spinal map prepared by the software and transferred into the intraoperative SpineAssist workstation. After the virtual template for the desired instrumentation has been created, a secondary verification procedure is performed, where tracked K-wire inserted into the mounted robot act to cross verify the accuracy of the template system. This secondary verification process assures an accuracy of less than 1.5 mm of deviation from the actual implant and the preoperative template. ⁷

The terminal point of registration occurs by obtaining six static fluoroscopic images utilized for calibration and registration. After registration and calibration have been completed, the SpineAssist software selects the optimal position of the selected arm for insertion of the drill sleeve and a cannulated drill guide is inserted into the arm which is automatically aligned along the predetermined trajectory. The drill then creates a cortical breach at the designated entry point and a guide wire is inserted into the vertebral body to allow a screw pilot hole to be drilled along the trajectory of the guide wire. A screw with an ideal length and diameter is then inserted into the pilot hole after pedicle probing and final confirmation of accuracy by the surgeon. ^{7 ,​ 8}

Early cadaveric studies of this novel robotic-assisted technique have reported average deviations from preoperative templates to actual implant position to be 1 mm or less. ^{6 ,​ 8} Togawa et al sought to evaluate the accuracy of bone-mounted miniature robotic systems for the percutaneous placement of pedicle and translaminar facet screws in a cadaveric series including 35 instrumented spinal levels. Using reconstructed 3D virtual X-rays of each vertebra, preoperative optimal entry points, and trajectories for screws were templated. A miniature robot was then mounted on to a clamp and the robotic system controlled the cannulated drill guide along the planned trajectory. K-wires were then advanced through the same cannulated guide trajectory and remained in the cadaver. Twenty-nine of 32 K-wires were found to be placed less than 1.5 mm of deviation, thus verifying the system’s accuracy and lending support for its use in MIS surgery. ⁶

Several clinical studies have attempted to determine the accuracy and efficacy of the SpineAssist robot (MAZOR Robotics Inc) in the clinical setting. ^{9 ,​ 10} Roser et al demonstrated a 99% accuracy rate for lumbosacral pedicle screw instrumentation when utilizing the SpineAssist robot compared to 98 and 92% utilizing fluoroscopic and navigation techniques, respectively. ¹¹ Utilizing the SpineAssist system, Schizas et al demonstrated a 95% accuracy rate in robot-assisted lumbosacral pedicle screw instrumentation compared to 92% accuracy rate with conventional fluoroscopy. ¹² Kantelhardt et al demonstrated similar results with a 95% accuracy versus 92% accuracy when utilizing SpineAssist compared to conventional fluoroscopy.

Alternatively, some studies have found decreased screw accuracy rates. A randomized controlled trial by Ringel et al found a diminished accuracy rate of lumbosacral pedicle screw instrumentation with SpineAssist robot at 85% compared to 93% in fluoroscopy-guided screws (p = 0.019). ¹⁰ The authors identified several points which may have led to their inferior robotic accuracy. The authors noted that utilization of a percutaneous means of fixation with one K-wire attached to the spinous process and two Steinmann pins inserted into the posterior superior iliac spines may have been insufficient. The authors also identified potential lateral deflection (slippage) of the cannula during screw insertion which could have contributed to the malpositioning of screws (which, when found, were laterally deviated). In turn, the authors have recommended superior fixation for the hover T K-wire (mounting system for Mazor) and utilization of a more lateral entry point, increasing screw medicalization and potentially avoiding “slippage” down the side of hypertrophic facets.