Workflow and Efficiency of Robotic-Assisted Navigation in Spine Surgery

Introduction

The increasing interest in improving intraoperative workflow and efficiency is evident from the growing body of literature on spinal robotics. The first robotic-assisted system for adult spine surgery received US Food and Drug Administration (FDA) clearance in 2004 [4]. Newer robotic systems with integrated surgical navigation have shown the potential for improved accuracy, shorter time-per-screw placement, less fluoroscopy/radiation time, and shorter hospital stays over freehand techniques [8,9]. One early iteration, the Mazor SpineAssist system, provided 6° of freedom of mobility for surgical instrumentation in addition to close to optimal screw positioning and placement [5]. Mazor Robotics released the second generation (Renaissance Guidance System) in 2011, which was followed by the third generation (Mazor X) in 2016. (Mazor Robotics was purchased by Medtronic in 2018.) The most recent version is the Mazor X Stealth Edition; first used in 2019, it combines navigation and robotics in a single platform [5]. Recent competitors include ExcelsiusGPS (Globus Medical, Inc., Audubon, PA, USA), ROSA Spine (Zimmer Biomet Robotics, formerly Medtech SA, Montpellier, France), and Cirq (BrainLab, Munich, Germany) [12].

While multiple studies have examined the operative metrics of robotic-assisted surgery, few have offered a practical understanding of the types of surgeries that can be performed using various systems and the intraoperative workflows required. Here we review these issues, with the aim of providing a foundation for surgeons seeking to incorporate robotics into their practice in an efficient manner.

We focus on the Mazor X Robotic Guidance System and the ExcelsiusGPS Robotic Navigation Platform; each has a different physical relationship to the operating room table. The Mazor X is a table-mounted system while the ExcelsiusGPS is a freestanding system with optical arrays. The Mazor X has an inherent stability due to its physical connection to the patient but, due to the weight of the bed-mounted arm, it cannot maintain that stability if the patient’s position changes. In addition, a 2 to 3 mm incision must be made on the patient’s right side for posterior superior iliac spine (PSIS) pin placement. In contrast, the ExcelsiusGPS is a freestanding system that does not attach to either the bed or the patient. This allows the patient to change positions without interfering with robotic arm stability.

The ROSA Spine received FDA clearance in 2016, and BrainLab’s Cirq received FDA clearance in 2019. Similar to the ExcelsiusGPS, ROSA is a freestanding system that uses a robotic arm and navigation camera, each mounted to mobile bases. BrainLab Cirq is similar to the Mazor X, as it is table-mounted.

Medtronic Mazor X Robotic Guidance System

Intraoperative Workflow

Once the patient is placed in the prone position on the Jackson table, a bed frame must be attached to the operating room table before transferring the patient. The robotic base should be placed at the foot of the table, as this will allow room for both the robotic arm mount and the C-arm. The C-arm can approach the patient from either side but should be on the side opposite the sterile field. The Mazor X Software Robotics image adaptor must be attached to the C-arm image intensifier prior to surgery.

Given the weight of the robotic arm, there is an on-board automated lifting mechanism stored within the robotic workstation that allows for mounting onto the bed frame. For more stability, the patient is directly mounted to the robotic arm (Fig. 1). There are different options for affixing to the patient’s bony anatomy, including use of the PSIS, a spinous process reference clamp, or a Schanz screw/pin into the pelvis. The Schanz pin is intended for open procedures from L4 to S2 or minimally invasive procedures from L1 to S2 [12]. When placing the Schanz pin, an incision is made directly over the PSIS. After placement of the pin, the Schanz screw ball adaptor will connect to the Schanz pin. This is then attached to the bone mount bridge. At this point, the robotic arm is linked to the patient’s anatomy and the surgeon may begin the process of registration. When mounting the robotic arm, it is typically placed at the foot of the bed on the patient’s right side. After the robotic system is mounted to the bed, it can be draped.

Fig. 1.

Mazor X Robotic Guidance System: (a) Setup for robotic arm and (b) the robotic arm is physically attached to the patient for more stability.

As a skin incision is made, it is important to remove all muscle and soft tissue off of the lamina. This step is to avoid any soft tissue pressure when the surgeon docks a cannula directly onto the bone surface. This allows for direct visualization and is an advantage for surgeons who are new to the system, as it helps them trust the technology. Additional preparation is often needed to smooth any uneven bony surfaces, to minimize skive potential. This should be done prior to the mounting process to minimize shift between the patient and the system once registered.

During registration, the Mazor X camera should be moved into a position where it recognizes the reference marker, so it can perform a 3-dimensional (3D) scan (Fig. 2a). It is important to note that before initiating this scan, blue towels should be placed over the surgical site and overhead lights turned away from surgical site (Fig. 2b). A “snapshot” is then performed to identify the location of the robotic arm for the navigation system. After that, anteroposterior (AP) and oblique fluoroscopy images are obtained with the surgical arm in both respective positions. This allows the system to calibrate the position of the arm in space. For AP imaging, a 3D marker is placed in the center of the AP fluoroscopy image. Once an AP image is obtained that meets these criteria, it is acquired by the software and the same steps are repeated for the oblique image.

Fig. 2.

Mazor X Robotic Guidance System: (a) During registration, the Mazor eye camera is moved to a position (top of figure) where it recognizes the reference marker and recognizes reference marker and will perform a 3D scan. (b) Blue towels are placed over the surgical site with lights removed before initiating this scan and (c) surgeon uses a Medtronic Powerease driver, which inserts the screw using a navigated screwdriver, aided by the saved trajectory plan. A virtual path as an exact computer-aided design model of the screw traversing the pedicle is projected onto the display and by using audible and tactile feedback.

These fluoroscopic images are then co-localized with the software planning template. During this step, all required surgical instrumentation is registered and verified. Registration is usually performed once; however, additional registrations may be required for longer instrumentation constructs [1].

Percutaneous Pedicle Screws: Wiltse Approach

Percutaneous pedicle screws are used for rigid internal fixation in minimally invasive surgeries and often require the use of Kirchner wires (K-wires) as a guide for screw insertion. The Wiltse approach is an alternative for precisely accessing and targeting pedicles, and incision sizes are roughly 4-mm long (Fig. 3). Incisions are made based on the number of levels and the convergent trajectory through pedicles into the vertebral body. Pilot holes are drilled with a burr within a navigated universal drill guide.

Fig. 3.

Mazor X Robotic Guidance System: Wiltse-skin incisions allow precise percutaneous access to the target pedicles. Incision sizes recorded in our cases reported as long as 4-mm long.

The trajectory plan is saved on Mazor X software prior to removal of the drill bit. Using the saved plan as a visual cue along with tactile feel, the pilot holes are then re-entered and tapped with an image-guided tap. This instrument is inserted using a low-speed/high-torque Powerase driver (Medtronic) that optimizes navigation accuracy by limiting the displacement of the spine associated with advancement through the pedicle (Fig. 2c). The screws are then inserted using a drill guide, aided by the saved trajectory plan [10]. Accurate screw placement is achieved by following the virtual path, while an exact computer-aided design model of the screw is projected onto the display. At this point, percutaneous rods are placed and any indicated compressions, distractions, and/or reductions are performed. A second fluoroscopy scan is taken to confirm accurate screw placement prior to the tightening of locking caps [11]. Subsequently, a 3D fluoroscopy scan is taken as an additional confirmation of accuracy.

Open Exposures: Midline Incision

In an open posterior fusion, a posterior midline skin incision is typically made followed by dissection of paravertebral muscles away from the lumbar spine. For each procedure, the robotic mount is attached to the PSIS or spinous process reference clamp, and fluoroscopic imaging is colocalized with the software planning template [1]. Of note, PSIS pins can be placed either before or after exposure. If the pin is placed prior to exposure, navigation can be employed during the dissection process, as the physical connection and registration will already have been achieved. One should always take care to avoid reference pin placement that obstructs planned screw trajectories, as this can lead to prolonged surgical time and an inability to place screws with robotic/navigation assistance. Avrumova et al reported events such as obstructive reference pin placement, which occurred in 0.6% of all robotically placed screws [1]. As a result, planned trajectories could not be achieved for certain pedicle screws.

It should be noted that “mini-open” procedures can also be performed robotically using separate fascial incisions. This approach avoids the larger incisions and extensive muscle dissection of traditional open surgery, and can contribute to faster patient recovery, reduced blood loss, and a reduced risk of infection [13].

Ziehm Scan

Following pedicle screw placement, patients may undergo an intraoperative scan with a 3D mobile C-arm (Ziehm Imaging GmBH, Nuremberg, Germany) [1]. C-arm images are taken intraoperatively before and after the procedure, in addition to 3D fluoroscopic imaging. This 3D imaging provides additional confidence in screw positioning, as implants can be evaluated in the axial, sagittal, and coronal planes [1]. Ziehm does not produce navigation or robotic platforms for spine surgery, but their scanners are compatible with certain imaging and navigation systems [10]. It is important to note that Ziehm imaging is not compatible with the Mazor X’s robotic navigation system, due to weight capacity issues. However, the Mazor X, of course, achieves co-localization between traditional intraoperative fluoroscopy and preoperative computerized tomography (CT) imaging.

Preoperative CT Scan Versus Intraoperative O-Arm Scan

The Mazor X system requires the patient to undergo a CT scan either preoperatively—sometimes called the CT-to-fluoro workflow—or intraoperatively (“scan and plan”). The CT-to-fluoro workflow requires a CT scan with 1-mm-thick slices, which is then loaded to the Mazor X software for planning purposes, prior to the incision [3]. It saves time in the operating room, as pedicle screw trajectories have been set beforehand [3]. As mentioned earlier, when using fluoroscopy, registration is typically preformed just once, unless a particularly long construct is planned [1]. As longer constructs are preformed, the field of view through fluoroscopic imaging is limited, therefore more than 1 registration would have to be done. However, this is an advantage of intraoperative CT (ICT) imaging with the O-arm. The O-arm has a larger field of view compared with fluoroscopic imaging, thus eliminating the restrictions and limitations for longer constructs and making 1 registration sufficient.

Globus ExcelsiusGPS System

Intraoperative Workflow

The workflow for the ExcelsiusGPS has some similarities and certain key differences when compared with the Mazor X. It provides 3 options with respect to imaging modalities: (1) preoperative CT with intraoperative fluoroscopy for registration, (2) ICT, and (3) intraoperative 3D fluoroscopy. In this section, we will describe the steps for intraoperative 3D fluoroscopy (Ziehm). Once the patient is placed in the prone position, the surgical tech registers the relevant instruments to a camera situated at the foot of the operating table. After the patient is draped, the procedure begins with placement of a single fiducial in the left PSIS. This fiducial serves as a “surveillance marker” that can detect if and when the fiducial array is inadvertently moved. Then, a dynamic reference base (DRB) is placed in the right PSIS. This is followed by attachment of the ICT registration device to the DRB. The patient is then draped, in preparation for a spin with the Ziehm scanner (3D fluoroscopy), taking care not to move the ICT or DRB.

Next, AP and lateral X-rays are obtained with the Ziehm to confirm that the spinal regions of interest will be captured, and to ensure that the ICT is fully visible. The 3D spin is then performed, after which the imaging data are transmitted to the robotic system and automatically registered to the patient. The ICT is then removed from the DRB. At this point, screw planning can commence. A touchscreen interface is used to determine the dimensions, location, and trajectory of each individual screw (Fig. 4a). Of note, when the screws are to be placed percutaneously for minimally invasive procedures, this planning step allows the screwheads to be truly co-linear to facilitate rod passage. Screw sizes and angles can also be designed to maximize pedicle fill.

Fig. 4.

ExcelsiusGPS System: (a) The surgeon selects a particular screw plan and then advances the robotic arm to the appropriate position using the foot pedal. (b) The screw is being placed with real-time imaging feedback and (c) a subsequent screw is being placed once the robotic arm has been advanced to appropriate position. The surveillance marker and DRB can be seen in the middle panel. DRB = dynamic reference base.

After the planning stage is complete, the robotic system is draped and the arm is brought into the surgical field. The system has an indicator that notifies the surgeon once the arm can reach all of the planned screws. At this point, the robot is locked to the floor and the arm’s end-effector is registered. Each screw is now selected in turn on the touchscreen, and a foot pedal is used to move the robotic arm into the appropriate position for that screw. The surgeon then uses a burr, through the robotic arm, to access the given pedicle while using real-time image guidance to ensure that the planned trajectory is being matched. The same maneuver is then repeated with a powered drill, after which the screw itself is placed (Fig. 4b). Tactile feedback, in addition to the image guidance, is exploited during this process. Warnings are also displayed if there is a concern for DRB movement or deviation from any planned trajectories. Once 1 screw is placed, the next screw is selected, and the foot pedal is again used to advance the arm to the appropriate position (Fig. 4c).

Next, the Ziehm is used to obtain AP and lateral fluoroscopic images for confirmation of appropriate positioning. Additional confidence in screw positioning is provided through 3D imaging, as the implants can be evaluated in the axial, sagittal, and coronal planes. Unlike the Mazor X system, the ExcelsiusGPS system is compatible with intraoperative Ziehm imaging, in addition to achieving co-localization between traditional preoperative CT imaging with intraoperative fluoroscopy and ICT imaging.

Preoperative CT With Intraoperative Fluoroscopy

With the ExcelsiusGPS system, the patient undergoes a preoperative CT scan, which is loaded into the system. The screw trajectories are planned out preoperatively in the sagittal, coronal, and axial planes. Intraoperatively, AP and lateral fluoroscopic images of the levels to be instrumented are obtained. Fluoroscopic images and the preoperative CT scan are then merged, and the surgeon verifies the reconstruction based on planned screw trajectories from fluoroscopic images and CT scans [6]. The robot is then draped sterilely, locked into place, and the robotic arm aligns the end effector to the planned trajectory; the desired screw label is selected. The screw hole is drilled and tapped, and the screw is then placed [6].

Intraoperative CT

Intraoperative CT scans can be obtained from a portable scanner such as an O-arm or a standard CT scanner at the time of surgery with the patient in the prone position [15]. After a CT scan is taken, the spinal levels are identified based on pedicle screw trajectories that were already planned and saved. Reference frames are installed and fixated to the pelvis, and instruments and arrays with reflective markers are registered [15]. The robotic arm is planned to pedicle trajectory, and incisions are made based on levels to be instrumented. Pedicle screws are inserted using navigated instruments guided by the robotic arm. Rods are placed, followed by locking caps. Intraoperative CT images are taken to verify screw and rod position [15].

The Learning Curve

Given that most surgeons lack familiarity with robotic systems, it is worth examining their learning curves. After all, any conclusion on the efficiency of a new technology must consider the initial adoption phase, during which procedures may not run smoothly. Kam et al studied robotic surgeries and found that robotic-assisted screw placement was highly accurate, with short instrumentation times and low complication rates. There was no difference in operative times between the initial adoption phase and later periods, which would imply a negligible learning curve at best [7]. Bäcker et al found that overall procedural times decreased as more robotic-assisted procedures were performed but also concluded that there was no major learning curve for the technology [2]. Interestingly, Urakov et al examined robotic surgeries and identified a trend toward improved efficiency with pedicle screw placement over time, although no statistically significant pattern was discovered [14]. Based on these studies, it reappears that the learning curve for robotic-assisted pedicle screw placement may be marginal.

In conclusion, navigated spinal robotics is a rapidly advancing technology. Concerns regarding the learning curve for surgeons, as well as for the entire operating room team, may be slowing adoption. Benefits such as improved screw accuracy and reduced radiation exposure may over time outweigh concerns that operative efficiency will be decreased.