Augmented reality-enabled image-guided surgical navigation in minimally invasive spine surgery: application beyond screw placement in prone lateral transpsoas interbody fusion—a cadaveric surgical technique study

Abstract

Augmented reality-enabled image-guided surgical navigation (ARE-IGSN) has demonstrated high accuracy for pedicle screw placement in spine surgery, but its application to interbody phases of minimally invasive fusion remains largely unexplored. Interbody work typically requires repeated fluoroscopic confirmation, increasing radiation exposure and forcing frequent gaze shifts away from the operative field. This study aims to evaluate the feasibility and accuracy of ARE-IGSN for key interbody steps in prone lateral transpsoas interbody fusion, including skin marking, corridor selection, disc preparation, evaluation of endplate integrity, and cage trajectory and depth control. A fresh frozen cadaver specimen was positioned in a prone lateral configuration with pelvic reference frame registration to a preprocedural computed tomography (CT) scan. An ARE-IGSN platform was used to provide in-field navigation overlays throughout the procedure. The workflow included navigated skin marking, docking, disc preparation using calibrated instruments, and interbody cage placement. Primary endpoints were technical feasibility, implant placement accuracy assessed by comparison of ARE-IGSN data with postprocedural CT imaging, and endplate integrity. Secondary endpoints included fluoroscopy utilization and qualitative ergonomic assessment. ARE-IGSN guidance was used throughout the procedure for skin marking, corridor selection, docking, disc preparation, endplate evaluation, trialing, and cage placement. The procedure was completed without the use of procedural fluoroscopy following system registration. Postprocedural CT demonstrated close agreement between the implant position displayed by the ARE-IGSN system and the final implant position. Retrospective comparison of measurements obtained from the saved ARE-IGSN data and postprocedural CT imaging demonstrated less than 1 mm difference across all assessed parameters, with no evidence of endplate violation. Visualization of navigational information was provided directly through the headset display throughout the procedure. This cadaveric study demonstrates the technical feasibility of applying ARE-IGSN guidance to multiple phases of prone lateral transpsoas interbody fusion beyond pedicle screw placement. Retrospective comparison of saved ARE-IGSN navigational data with postprocedural CT imaging demonstrated close agreement between the implant position displayed by the system and the final implant position. These findings support further prospective clinical investigation of implant placement accuracy, workflow integration, radiation utilization, and ergonomic outcomes.

Highlight box.

Surgical highlights

• Navigated skin marking and corridor selection are performed using ARE-IGSN overlays. Instrument docking is guided by in-field visualization of safe zones and exclusion boundaries. Disc preparation performed with calibrated instruments showing real-time proximity to endplates to avoid violation. Cage trajectory and depth relative to the planned implant position established during surgical planning using ARE-IGSN without procedural fluoroscopy.

What is conventional and what is novel/modified?

• Conventional prone lateral interbody fusion relies on repeated intra-operative two-dimensional fluoroscopic radiographs with guidance viewed on a remote monitor.

• This technique introduces the concept of replacing traditional fluoroscopic imaging with ARE-IGSN in-field navigation throughout the entire workflow. Continuous head-up visualization eliminates attention shifts away from the operative field to remote monitors.

What is the implication, and what should change now?

• ARE-IGSN has the potential to extend navigation beyond pedicle screws to interbody phases, influencing surgical success with a more efficient workflow that minimizes radiation exposure.

• This pilot cadaveric feasibility study demonstrates technical viability and defines measurable endpoints for future clinical application.

Introduction

Background

Augmented reality-enabled image-guided surgical navigation (ARE-IGSN) platforms integrate preoperative and/or intraoperative medical imaging with patient-specific anatomical models, image registration, optical or electromagnetic instrument tracking, and augmented reality (AR) visualization to provide real-time surgical guidance. Following registration of the patient’s anatomy to imaging data, three-dimensional holographic representations of relevant anatomical structures, planned procedural pathways, and tracked surgical instruments are spatially overlaid onto the operative field using a head-mounted display or equivalent AR visualization device.

Unlike conventional image-guided navigation systems, which require surgeons to repeatedly alternate their attention between the patient and an external navigation monitor, ARE-IGSN systems present navigation information directly within the surgeon’s field of view. This enables continuous visualization of patient-specific anatomy, instrument position, procedural targets, and planned trajectories while maintaining visual focus on the operative field, thereby enhancing spatial awareness and hand-eye coordination throughout the procedure.

Across prospective and matched studies, ARE-IGSN guidance has produced high pedicle screw accuracy, including a matched cohort in which clinically accurate screws by Gertzbein-Robbins grading were more frequent with ARE-IGSN than with freehand placement (93.9% versus 89.6%). In a consecutive clinical series using a head-mounted display, the overall Grade A or B accuracy approached 98%, supporting reliability in routine use (1,2). Several ARE-IGSN platforms are cleared by the Food and Drug Administration (FDA) for intraoperative spine applications and are increasingly used as a navigation modality within standard practice (3). As evidence continues to accumulate regarding accuracy, workflow efficiency, reduction in radiation exposure, and educational utility, AR is expected to play an expanding role in image-guided spine surgery (4).

Minimally invasive interbody fusion relies heavily on repeated fluoroscopic confirmation for level localization, skin marking, instrument docking, discectomy, endplate preparation, and assessment of cage trajectory and depth (5). This repeated imaging exposes both the patient and the operative team to ionizing radiation and forces the surgeon to alternate their gaze between a distant monitor and the field with associated head, neck, back, and shoulder movements (6). Occupational exposure during common spine fluoroscopic projections is well documented and can be substantial for staff positioned near the radiation source (7). Even when protective shielding is used, cumulative exposure over a surgeon’s career remains a concern, and radiation risk extends to all members of the operative team, including assistants, scrub technicians, and circulating staff.

Rationale

When navigation is anchored to a three-dimensional acquisition, ARE-IGSN systems, in particular those that allow for AR visualization through an AR headset and not a standard monitor, shifts confirmations into the field of view and reduce or eliminate the need for fluoroscopy use (8). In a systematic review, ARE-IGSN with cone-beam or computed tomography (CT)-based registration was associated with a lower dose area product than fluoroscopy-guided freehand workflows. Staff exposure in ARE-IGSN guided cases has been reported at approximately 0.21 microsieverts per case when team members are shielded during the single three-dimensional acquisition (9). Beyond radiation, presenting guidance in the same line of sight reduces attention shifts and head turns that occur with conventional systems, a human factors advantage that has been measured in image-guided settings and highlighted in reviews of ARE-IGSN for minimally invasive spine surgery (10). This head-up paradigm also supports smoother procedural flow and allows the operative team to anticipate key steps based on visible trajectory and target cues, potentially improving coordination and efficiency.

Most published applications of ARE-IGSN in spine surgery have focused on pedicle screw placement. However, interbody work in minimally invasive fusion is also dependent on accurate trajectory control, depth control, and preservation of endplate integrity. Extending ARE-IGSN guidance to these phases may provide additional value beyond instrumentation alone.

Objective

The objective of this study is to evaluate the feasibility of extending the use of ARE-IGSN beyond screw placement (where AR visualization is delivered via a head-mounted display) during a cadaveric minimally invasive prone lateral transpsoas interbody fusion procedure. In this study, we used the Medivis Spine Navigation system built on the SurgicalAR spatial computing platform (Medivis, Inc., New York, NY, USA). We used in-field overlays to guide skin marking, corridor selection, disc preparation, endplate evaluation, and cage placement. The study assessed technical feasibility, implant placement accuracy, and preservation of endplate integrity. In addition, we documented fluoroscopy utilization and qualitative ergonomic observations to help inform future prospective investigations. We describe a minimally invasive prone lateral transpsoas surgical technique in a cadaveric model and present this article in accordance with the SUPER reporting checklist (available at https://jss.amegroups.com/article/view/10.21037/jss-2026-0084/rc).

Preoperative preparations and requirements

Setting and collaborators

This study was performed during a single fresh-frozen cadaver session at the inaugural Houston Methodist Bookout Center Surgical Technology and Robotics Summit (STaRS). Participants included a board-certified spine surgeon, a spine surgery clinical fellow, Bookout Center and Houston Methodist Institute for Technology, Innovation, and Education (MITIE) staff, representatives from Medivis for platform support, our research team, and an ATEC Spine (Carlsbad, CA, USA) representative. ATEC Spine provided lateral access instruments, trials, and an interbody device. The Bookout Center served as the on-site point of contact for setup and troubleshooting with industry partners.

Compliance

The study was conducted under institutional body donation policies and MITIE laboratory governance. No live human subjects were involved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. All cadaveric procedures were approved, performed, and managed in accordance with the policies and procedures of the Texas Funeral Service Commission, as well as all applicable institutional and regulatory requirements. The study did not require approval from the Institutional Review Board of Houston Methodist. The Medivis Spine Navigation system platform is FDA-cleared for open and minimally invasive spine surgery. Radiation safety practices were followed for CT image acquisition.

The concept and design of the study were developed independently by the authors. Collaboration with industry partners was facilitated through the Houston Methodist Bookout Center as part of the Houston Methodist Academic Institute and conducted in accordance with institutional governance, policies, and regulatory standards.

Specimen and positioning

One full cadaveric specimen with a lumbar target level was used. The specimen was positioned in a prone lateral configuration on a Jackson table (Mizuho OSI, Union City, CA, USA) with standard padding. A pelvic pin at the posterior superior iliac spine hosted the patient reference frame.

Imaging and registration plan

A CT scan was obtained (Siemens Healthineers AG, Erlangen, Germany) and uploaded to the ARE-IGSN system. Registration was performed to the pelvic reference and verified. The headset remained on throughout and did not interfere with communication or instrument handling.

Objectives

The objectives were to demonstrate the feasibility of ARE-IGSN for navigated skin marking, corridor selection, and docking disc preparation with endplate awareness, cage trajectory, and depth control. Screws were planned for construct realism but were not the primary focus since accuracy for screws with ARE-IGSN is already established in the literature.

Step-by-step description

Lateral skin marking and corridor selection

Skin marking for a single-level prone lateral approach was performed with a navigated probe using ARE-IGSN overlays rather than iterative fluoroscopy. The overlay displayed level confirmation and a side lock that fixed guidance to the intended approach side, a clear visual corridor to the disc space, and explicit exclusion zones (anterior longitudinal ligament, major vessels, lumbar plexus, and spinal canal). The osseous anatomy remains clearly visualized on the overlay which includes the vertebral body boundaries, pedicles, transverse processes, facet joints, and endplate orientation, allowing precise localization of the disc space while avoiding encroachment on the posterior elements and adjacent bony structures (Figure 1A).

Figure 1.

ARE-IGSN-assisted prone lateral workflow. (A) Skin marking with a navigated probe and in-field overlays that show level confirmation, side lock, the corridor to the disc, and exclusion zones for the anterior longitudinal ligament, major vessels, the lumbar plexus, and the spinal canal. (B) Retroperitoneal access and docking with disc level confirmation using the probe and headset display. (C,D) Disc preparation using calibrated navigated instruments while endplate boundaries and proximity cues are visible. (E,F) Trialing and cage delivery along the planned trajectory with target depth cues. (G-I) Posterior steps for context only including screw planning on the scan, preliminary screw placement, rod seating, and set screw tightening. These images are published with the participant’s consent. ARE-IGSN, augmented reality-enabled image-guided surgical navigation.

Retroperitoneal access and docking

Standard retroperitoneal lateral access was performed using blunt finger dissection in a controlled, stepwise manner to develop the retroperitoneal space and sweep the peritoneal contents anteriorly, as per established technique. The headset display did not impact this step. Once within the retroperitoneal corridor, the target disc space was identified by digital palpation of the lateral vertebral bodies and disc contour and confirmed with the navigated probe. Sequential dilators were then advanced to the disc space along the planned trajectory, maintaining orthogonal alignment to the disc plane. No neuromonitoring was used in this cadaveric model. Navigation accuracy was confirmed through alignment of anatomical landmarks with augmented overlays. The retractor system was subsequently deployed at the predetermined entry point. Final positioning within the disc space, including anterior-posterior orientation, was verified by tactile feedback and confirmed with the navigated probe using the in-line augmented display (Figure 1B).

Disc preparation and endplate protection

Navigated instruments, including Cobb elevators, curettes, and ring curettes, were calibrated to the navigation system. Disc preparation proceeded with the headset showing endplate boundaries and proximity cues as the working tip approached predefined limits. The corridor remained in view, and no gaze shift to a remote monitor was needed (Figure 1C,1D).

Trialing and cage delivery

Trials were calibrated and advanced until a stable endplate fit was achieved. A polyetheretherketone (PEEK) interbody device (IdentiTi™-LIF; ATEC Spine) was selected and delivered along the planned trajectory under navigated augmented visualization. The implant is composed of radiolucent PEEK with radiographic markers to aid intraoperative localization and is available in a range of footprints and lordotic profiles to accommodate patient anatomy. In this case, a 55 mm × 18 mm footprint cage was utilized. The implant was advanced in line with the disc space, with target depth cues visible in the headset. Final positioning appeared to match the planned implant position displayed within the AR environment. Quantitative assessment of implant position relative to surrounding osseous landmarks was recorded, including its relationship to the anterior and posterior vertebral body margins and the superior and inferior endplates in both the anterior-posterior and medial-lateral planes (Figure 1E,1F).

Posterior instrumentation

We planned screw vectors and sizes on the operative scan and introduced preliminary screws to maintain construct realism, then seated the rods and tightened the set screws after the interbody work was completed. These posterior steps are documented for completeness only. They are not part of the study outcomes, and detailed placement mechanics are omitted since the accuracy of AR-guided pedicle screws has already been established in prior literature. (Figure 1G-1I).

Operative time

Operative time was not formally recorded in this cadaveric feasibility session, as the primary objective was technical validation of workflow and accuracy. Future studies will include structured time measurements to assess procedural efficiency and reproducibility.

Postoperative considerations and tasks

Verification imaging and outcomes

Implant placement accuracy was assessed retrospectively by comparing the implant position displayed within the ARE-IGSN headset environment during the procedure with the final implant position on postprocedural CT imaging.

Following completion of the procedure, navigational data recorded and stored within the ARE-IGSN platform were reviewed retrospectively. Using the measurement tools available within the software, measurements were obtained from the displayed implant position to the anterior vertebral body cortex, posterior vertebral body cortex, and the right and left lateral vertebral body cortices on axial imaging. A verification CT scan was then reviewed and the same measurements were repeated manually using picture archiving and communication system (PACS) measurement tools on the postprocedural CT images (Figure 2).

Figure 2.

Verification imaging and outcomes. (A) Sagittal CT demonstrating final implant position with preserved endplate contours. (B) Coronal CT confirming appropriate implant depth and medial to lateral position with cortical support and no cortical breach of the vertebral endplate margins. (C) Axial CT demonstrating implant position consistent with the position displayed by the augmented reality system during the procedure. CT, computed tomography.

Differences between the implant position displayed by the ARE-IGSN system and the final implant position on CT were calculated by comparing corresponding measurements. Retrospective comparison of measurements obtained from the saved ARE-IGSN data and postprocedural CT imaging demonstrated less than 1 mm difference between corresponding measurements relative to the anterior cortex, posterior cortex, and right and left lateral vertebral body cortices.

The final implant position remained well contained within the vertebral body margins, and no endplate violation was identified. Pedicle screw parameters were consistent with the registered trajectories displayed by the system.

All measurements were performed retrospectively by a member of the study team (A.A.) after completion of the procedure and were intended as a feasibility assessment rather than a formal validation study. Independent measurement verification was not performed.

Radiation and ergonomics

The session was completed without the use of procedural fluoroscopy following registration. Visualization of navigational information was provided directly through the ARE-IGSN headset display throughout the procedure. Observations regarding surgeon posture and head movement were qualitative in nature and were not formally measured. Accordingly, no conclusions regarding ergonomic benefit can be drawn from this cadaveric feasibility study.

Tips and pearls

Accurate registration to a rigid pelvic reference frame and verification of navigation accuracy before each critical step are essential for reliable AR guidance. During docking, combine navigated overlays with tactile confirmation of the disc space rather than relying solely on virtual visualization. Recalibrate navigated instruments after exchange to maintain tracking accuracy throughout disc preparation. Use the in-field endplate and trajectory overlays to preserve orthogonal instrument alignment and avoid endplate violation during preparation and cage insertion. Finally, perform a final navigated assessment of implant trajectory and depth before cage impaction to confirm satisfactory positioning without additional fluoroscopic imaging.

Discussion

This cadaveric feasibility study demonstrates the technical feasibility of applying ARE-IGSN to multiple phases of a prone transpsoas interbody workflow, including navigated skin marking, corridor selection, docking, disc preparation, trialling, and cage delivery. The procedure was completed without procedural fluoroscopy following registration, and retrospective comparison of saved navigational data with postprocedural CT imaging demonstrated close agreement between the implant position displayed by the ARE-IGSN system and the final implant position. These findings should be interpreted as proof-of-concept observations rather than evidence of clinical superiority or improved patient outcomes.

The present findings should be interpreted in the context of prior clinical work on AR in spine surgery. Carl et al. reported clinical implementation of microscope-based AR support in spinal surgery, demonstrating integration of patient specific anatomy, tumors, spinal cord structures, and instrumentation into the operative field to improve spatial understanding and intraoperative orientation (11,12). Similarly, Sommer et al. described a clinical AR workflow for minimally invasive transforaminal lumbar interbody fusion in which sequential anatomical landmarks were displayed throughout the procedure to facilitate surgical approach, facet resection, disc space identification, cage placement, and contralateral decompression (13). These studies provide important clinical evidence supporting the feasibility of AR in spine surgery beyond pedicle screw placement alone.

The present study differs from these prior reports in several important ways. First, we focused specifically on a prone lateral transpsoas interbody workflow rather than posterior or transforaminal approaches. Second, the AR platform was used throughout transpsoas corridor development, disc preparation, endplate evaluation, trialing, and cage delivery using a head-mounted display platform. Finally, we retrospectively compared the implant position displayed by the AR system with the final implant position on postprocedural CT imaging. Accordingly, our findings should be viewed as a focused feasibility extension of the existing AR spine literature rather than evidence of clinical superiority.

The practical value of ARE-IGSN in this context is that navigational information is displayed directly within the surgeon’s field of view via a headset rather than on a separate monitor. In this session, skin marking, docking, disc preparation, trialing, and cage delivery were performed using in-field visualization provided by the headset. In addition, the system displayed anatomical boundaries and implant position information throughout the procedure. The ability to visualize this information continuously may have implications for workflow and ergonomics; however, these outcomes were not formally measured and require prospective evaluation.

Postprocedural CT imaging demonstrated close agreement between the implant position displayed by the ARE-IGSN system and the final implant position. Assessment was performed retrospectively using saved navigational data and verification imaging. Differences between corresponding measurements were less than 1 mm across all assessed parameters. These findings support the technical feasibility of the workflow but should not be interpreted as formal validation of implant placement accuracy, as measurements were performed retrospectively by a single investigator and were not independently verified.

This experiment also highlights practical endpoints that can be assessed in future clinical studies. Although the procedure was completed without procedural fluoroscopy following registration, no comparator workflow was performed, and no formal radiation analysis was undertaken. Similarly, ergonomic observations were qualitative and were not objectively measured. In the operating room, a prospective comparison should capture machine dose area product and personal badge readings for the surgeon, assistants, scrub nurse, and circulating nurse, record fluoroscopy time and number of images when used, and track the proportion of confirmations obtained through ARE-IGSN versus fluoroscopy. Human factors should be documented with head turn counts, a short workload rating such as National Aeronautics and Space Administration Task Load Index (NASA TLX), a brief posture snapshot such as Rapid Upper Limb Assessment (RULA) or a single sensor measure, and a note of any headset-related discomfort. These measures are simple to collect and align directly with the proposed benefits of in-field visualization and reduced attention switching.

The session exposed several engineering targets to pursue. Manual calibration of navigated instruments was reliable but added steps and could be a source of small registration drift over time. A curated library of pre-recognized instruments with quick checks would streamline setup and would likely improve consistency during disc work where multiple tools are used. Boundary visualization for endplate limits and for the lateral edge can be tuned for contrast and depth perception, and depth cues during cage delivery can be enhanced for users who prefer a more explicit arrival signal at the planned landmark. The headset video and synchronized audio, combined with event logs and time stamps, will allow the engineering team to link observed challenges to specific moments and will help prioritize development.

It is essential to acknowledge the constraints of a cadaver model. There is no live nerve function, so we could not confirm clearance from the lumbar plexus with triggered stimulation through dilators or retractors, and we could not detect transient irritability during docking or disc manipulation. In this session, corridor safety was inferred from anatomy and from the augmented display without neurophysiologic feedback. Additionally, the absence of tissue turgor, bleeding, and respiratory motion simplifies registration and reduces corridor drift, which can overstate stability and ease compared with a live case. For that reason, our results should be interpreted as feasibility signals that define a protocol and a set of endpoints rather than as evidence of clinical superiority.

Taken together, these findings suggest that ARE-IGSN may be applied to phases of interbody work beyond pedicle screw placement. In this cadaveric session, retrospective comparison of saved navigational data with postprocedural CT imaging demonstrated close agreement between the implant position displayed by the ARE-IGSN system and the final implant position. Future prospective clinical studies should evaluate implant placement accuracy, endplate integrity, fluoroscopy utilization, operative time, neuromonitoring integration, and ergonomic outcomes using predefined assessment criteria.

Limitations

This was a single cadaver session in a controlled laboratory setting without physiologic motion or bleeding, which can underestimate intraoperative complexity. The specimen does not provide live nerve function and therefore proximity to the lumbar plexus could not be assessed with triggered electromyography through dilators or retractors. In a live case, the team would test and stimulate instruments to ensure clearance from neural elements during docking and disc work. In this cadaver experiment, corridor safety was inferred from anatomy and augmented overlays without neurophysiologic feedback. The platform used did not include a preassigned library of navigated instruments, and each tool required manual calibration, which added procedural steps and may introduce small degrees of registration drift. No quantitative radiation comparison was possible, as no control workflow was performed, and ergonomic observations were qualitative in nature. Operative time was not recorded, limiting assessment of workflow efficiency. These findings should therefore be interpreted as feasibility signals that motivate structured clinical study.

Future directions

Future studies should evaluate ARE-IGSN interbody workflows in clinical settings with prospective assessment of implant placement accuracy, endplate integrity, fluoroscopy utilization, operative time, radiation exposure, and ergonomic metrics. Additional investigation is needed to assess system performance under physiologic conditions, integration with neuromonitoring, and workflow reproducibility across different surgeons and institutions. Comparative studies against fluoroscopy-guided and conventional navigation-assisted workflows will be important to better define the clinical role of ARE-IGSN in minimally invasive spine surgery.

From a technical perspective, future development should focus on integration of pre-calibrated navigated instrument libraries, reduction of manual calibration requirements, improvements in registration robustness, and enhanced real-time feedback during disc preparation and implant delivery.

Implementation and cost considerations

Implementation of ARE-IGSN workflows requires specialized hardware, imaging infrastructure, and personnel familiar with system setup and operation. Initial costs include acquisition of navigation platforms, integration with imaging systems, and training requirements. Although ARE-IGSN may reduce reliance on procedural fluoroscopy and provide in-field visualization of navigational information, the cost-effectiveness of these systems remains uncertain and requires prospective evaluation. Further studies are needed to determine scalability, resource utilization, and economic impact across different practice environments.

Conclusions

In a single cadaveric prone transpsoas session, ARE-IGSN was successfully applied to navigated skin marking, corridor selection, disc preparation, endplate evaluation, and cage delivery. Retrospective comparison of saved ARE-IGSN data with postprocedural CT imaging demonstrated close agreement between the implant position displayed by the system and the final implant position. These findings support the technical feasibility of extending ARE-IGSN beyond pedicle screw placement and provide a foundation for prospective clinical evaluation.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. All cadaveric procedures were approved, performed, and managed in accordance with the policies and procedures of the Texas Funeral Service Commission, as well as all applicable institutional and regulatory requirements. The study did not require approval from the Institutional Review Board of Houston Methodist.