2.1. Workflow

Our workflow included the following: preoperative imagining, screw planning, intraoperative imaging with automatic registration, fusion of the preoperative and intraoperative imaging with a review of the preplanned screw trajectories, robotic-assisted insertion of K-wires followed by fluoroscopy-assisted insertion of pedicle screws and a control iCT scan.

Prior to surgery, all patients received a 3D CT of the region of the spine that was planned for instrumentation. This dataset was exported to the navigation software (BrainLab, Munich, Germany). Using the crew planning application, the radius, length and trajectory of the pedicle screws were planned prior to surgery (Figure 1). The software automatically recognizes the vertebras which are planned for instrumentation and automatically sets the screws. In the next step, the screw direction, radius, and length were adjusted manually. After the trajectory was planned, a review of the trajectory was performed with final adoption.

Preoperative screw planning using Screw planning application (BrainLab, Munich, Germany).

The intraoperative setting with iCT has been described elsewhere [14]. The patient was placed on a mobile, radiolucent carbon-fiber surgical table (TruSystem^® 7500, Trumpf Medical Ditzinger, Germany) for spinal applications, connected to the CT scanner (AIRO^®, Brainlab, Munich, Germany). Surgery was performed in the scanning position with anesthesia cables and lines routed through the gantry. Cirq^® was attached and draped on the left side of the operating table on a metal holder. (Figure 2.)

Perioperative setting. (A). Cirq^® is attached on the left side of the operating table. (B). Surgical field is prepped and draped. Four fiducials are attached to the skin prior for accuracy check. Reference array is in this case of percutaneous, minimally invasive pedicle screw implantation, is attached to a spinous process proximal to the surgical field via separate skin incision. (C). View of the patient while performing initial registration iCT scan.

The correct spinal levels planned for the instrumentation were determined by fluoroscopy for planning the skin incision. Four artificial adhesive skin fiducials were placed on both sides of the planned incision to monitor accuracy.

The surgical field was prepped and draped. A carbon reference array was attached to the spinous process in open surgery, with one segment cranial to the cranial end of the surgical field. For percutaneous, minimally invasive surgical cases, a small midline incision cranial to the surgical field was performed, with a subperiosteal preparation of the spinous process, to allow for the fixation of the carbon reference array to the spinous process.

In cases of the stabilization of multiple levels, the reference array was sometimes moved closer to the instrumented vertebra for better accuracy. For example, in cases for which stabilization Th10-L2 was performed, the carbon reference array was initially placed at Th9. Following the registration scan, robotic-assisted implantation of K-wires was performed in Th10, 11 and 12. Pedicle screws were then implanted under fluoroscopy control. The reference array was then moved to the spinous process of Th12, and a control CT scan was performed, covering the area of Th10-L2. This scan was used for implant-position control as alongside a registration scan for the stabilization of L1 and L2.

A sterile coverage was applied on the patient, so that the reference array was still visible by the navigation camera. Following this, a navigated low-dose intraoperative CT scan, covering the region of interest (spinal segments which were planned for the stabilization), was performed. Immediately after scanning, imaging data was automatically transferred to the navigation system (BrainLab, Munich, Germany) without user interaction for automatic patient registration with the acquired image data [14].

To display preplanned screw trajectories in the recent imaging data, the co-registration of preoperative data, defining a region of interest with pre-planned screws, was conducted. Elements Spine Curvature Correction co-registers scans of the patient to compensate for inevitably varied spine positions during different imaging sessions. The software brings together scans from preoperative CT with iCT to update them for surgery. Additionally, if needed, preoperative magnetic resonance imaging (MRI) of the spine can be fused with preoperative and intraoperative imaging. Preoperative image data from CT and MRI are fused non-linearly, by applying the spine curvature element in a rigid and elastic fashion (rigid Elements Image Fusion 3.0 or elastic Elements Curvature Correction, Brainlab, Munich, Germany) [15].(Figure 3). Rigid fusion was only used in the region of segments which were planned for mono- and bi-segmental stabilization. Elastic fusion included the registration of the pre-aligned data and was used alongside rigid fusion in cases of multisegmented stabilization and for thoracolumbar constructs.

(A). Rigid fusion of the preoperative CT of the lumbar spine and registration iCT scan with preplanned screw trajectories (Patient number 3). (B). Elastic fusion of the preoperative CT of the lumbar spine and control iCT in the case of multisegmented stabilization in the thoracolumbar spine (Patient number 2).

Registration accuracy was evaluated by placing the pointer tip on anatomical landmarks such as the spinous process, vertebra lamina, or the vertebral body surface, or on artificial landmarks, such as skin fiducials or mini-screws attached to the spinous process [15]. A registration accuracy check can also be performed following control iCT with the pointer tip placed on the implants (Figure 4). For calculation of the effective dose (ED), the total dose length product (DLP) was multiplied by ED/DLP conversion factors, which are estimated to be 17.8 μSv/Gy × cm for thoracic, and 19.8 μSv/Gy × cm for lumbar scans [16,17]. The DLP refers to a phantom with a diameter of 32 cm for thoracic and lumbar examinations.

Registration accuracy check with tip of the pointer on (A,B) the skin fiducials (C). holder of the carbon reference array (D). the mini screw attached to the spinous process. (E–G). head of the screw.

An intraoperative CT was used for the robot-guided implantation of K-wires into the pedicles following the insertion of pedicle screws under fluoroscopy control.

The Cirq^® Robotic Alignment Module consists of an interchangeable head, which enables alignment with the preplanned trajectory and a kinematic unit for fine-tuning adjustment, based on pre-planned trajectories, during computer-assisted robotic surgery. A tracking array was attached to the kinematic unit for real-time tracking of the instruments and constant position feedback (Figure 5). The first step is to move the robotic arm over the entry point of the preplanned trajectory. When the robotic arm has been roughly placed over the entry point of the trajectory, the robotic alignment module enables the automatic positioning of the robotic arm according to preplanned screw trajectory. A drill guide is then inserted through the tracking array and positioned onto the entry point; following this, the tracking array can be safely locked, which further secures the drill guide. Snap-on depth control is attached to the proximal part of the drill guide for precise and safe drilling. Following drilling, a K-wire is implanted, and the robotic arm detached. (Supplemental Material: Video). Here, the surgeon stood on the left side of the patient, next to the robotic arm (Figure 6).

CIRQ Robotic Alignment Module. (A). Kinematic unit (arrow), tracking array (double arrow). (B). Intraoperative view. Attachment of drill guide (star) onto the kinematic unit. Reference array attached to the spinous process (triple arrow).

Position of the surgeon during the procedure. Reference array (arrow).

A CT scan was performed at the same time as the scan for the control of implanted pedicle screws and a registration scan for subsequent stabilization. Following the implantation of K-wires, pedicle screw implantation was performed under fluoroscopy control. After screw placement, an intraoperative CT scan was performed to confirm correct screw placement.

The following parameters for the assessment of the initial experience were used for evaluation:

Surgery time was defined as the period between the first incision and closure.

The time taken for positioning and robot installation was defined as the time necessary to position the patient and completely install the robotic arm.

The robotic time, which was recorded by the robotic system, was the total time in which the robotic arm was in use. This period covers the point of initiation of the first entry point search to the implantation of all K-wires.

Time per screw—since we did not measure time needed for the implantation of each screw separately, time per screw was calculated per case by dividing the robotic time with the number of implanted screws. This provides a measurement of the time needed for robotic-assisted implantation of K-wires. Since K-wire implantation is the essential robotic-supported element of pedicle screw implantation, we labeled this time as time per screw.

Pedicle size—measurement of the pedicle size in the axial, coronal and sagittal plane, on the registration iCT scan, with a line perpendicular to the preplanned screw

Pedicle screw accuracy—measured according to the CT-based Gertzbein and Robbins System (GRS). The grading system reflects the deviation of the screw from the “ideal” intrapedicular trajectory. The transpedicular screw position was graded from A to E, based on the extent to which the screw breaches the cortex of the pedicle 1–3:

entire intrapedicular position without a breach of the pedicle cortex

exceeding the pedicle cortex < 2 mm

exceeding the pedicle cortex 2–4 mm

exceeding the pedicle cortex 4–6 mm

exceeding the pedicle cortex > 6 mm or reaches outside of the pedicle.

Grade A and B can be considered as satisfactory operation results. In grade C to E, neurological symptoms may occur and therefore, can be evaluated as an unsatisfactory surgical result [18].

Deviation of preplanned trajectory from actual pedicle screw position—offset of the screw compared to preplanned trajectory (degrees to medial/lateral): mean deviation in entry point, the average deviation from the tip of the screw, and angular deviation. Deviations were determined using an image-overlay analysis to compare preoperative CT imaging with preplanned screw trajectories to a true screw position on intraoperative, control CT imaging (Figure 6). The mean deviation of entry point and the average offset from the tip of the screw were measured in the axial plane by determining the perpendicular distance of the midline of the planned screw versus the midline of the actual screw position; this latter line was drawn manually in the software as a best estimate on the slice with the widest screw diameter. Angular deviation was measured by determining the angle of the midline for the planned screw versus the midline of the actual screw position in the lateral plane.

Figure 7 is from the low dose iCT, these are the original resolution data.

Accuracy of the implanted pedicle screws and deviation compared to the preoperatively planned screw trajectories were measured using the image overlay technique following rigid and elastic fusion of the preoperative CT scan, registration iCT scan and control iCT scan. (A). Implanted screws are segmented in red and the trajectory of the preplanned screws in blue. Entry point and tip point deviation were measured in the axial plane. (B). Overlay of a preplanned screw trajectory with the implanted pedicle screw. Entry point deviation 3 mm to medial and lateral from the entry point of the preplanned screw trajectory was considered acceptable for correct pedicle screw placement. (C). Angular deviation was measured in the lateral plane via the angular measurement tool in the BrainLab software using the measurement of the angle between the axis of the implanted screw and the preplanned screw trajectory.