An immersive mixed reality surgical self-training system for precise pedicle screw placement: a randomized controlled trial

Abstract

Background

Given the long and costly training cycle required for pedicle screw placement, we proposed an immersive mixed reality surgical self-training system (IMR-SS) for pedicle screw placement. IMR-SS combines holographic real-time training instructions with a physical spine phantom to integrate virtual simulation with hands-on training. This study evaluated the effectiveness of IMR-SS in a randomized controlled trial. We aimed to determine whether IMR-SS improves novice learners’ pedicle screw placement performance and training satisfaction.

Methods

The IMR-SS for pedicle screwing consists of three parts: a teaching module, a hands-on module, and an assessment module. The IMR-SS integrates 3D-printed models, real surgical instruments, and immersive mixed reality technology to provide an immersive learning experience, high-fidelity haptic feedback, and real-time instructions. A randomized controlled trial was conducted with 32 undergraduate medical students from two centers. Participants were randomly assigned to either the IMR group (using IMR-SS) or the control group (using a digital textbook). Both groups underwent theoretical and practical training, followed by identical assessments.

Results

Compared with the control group, the IMR group presented significantly higher completion rates (0.99 ± 0.02 vs. 0.87 ± 0.11, p < 0.01) and fewer errors (0.06 ± 0.25 vs. 2.13 ± 1.54, p < 0.01). The IMR group showed better performance with significantly smaller screw placement angles (13.88°±6.98° vs. 20.89°±11.59°, p = 0.049). The theoretical assessment revealed no significant difference between the two groups, indicating equivalent baseline knowledge. Compared with the control group, the IMR group had greater training satisfaction and greater confidence in training outcomes.

Conclusions

The IMR group reported greater satisfaction with and confidence in the outcomes of screw placement training. The IMR-SS is a feasible and effective method for enhancing surgical operation education for novice medical students, providing superior hands-on training experiences and improving practical skills. Future research should focus on long-term learning curve validation and skill transferability and develop more surgery curricula for generalizability validation.

Introduction

The necessity of a prolonged training period to develop competent surgeons is a major hurdle that current orthopedic surgical training faces [1, 2]. The traditional model of “see one, do one, teach one” might not be capable of coping with the growing demands for surgical training and education [3, 4]. For pedicle screwing, approximately 80 screws need to be placed, and 25 patients are proficient as attending surgeons [5]. Cadaveric dissection has long been the gold standard for hands-on training, but cadavers present ethical issues, high costs, and limited availability [6, 7]. synthetic bone phantoms offer a cadaver-free alternative but are expensive and lack the anatomical variability of real patients [8, 9] Moreover, the increasing demand for academic outcomes and residency workloads also leads to fewer opportunities for hands-on practice and expert instruction [2, 10,11,12]. These challenges underscore the need for independent self-training systems that provide customizable resources and expert guidance to supplement traditional training.

To address these gaps, a variety of simulation-based tools have been explored for surgical education. Immersive virtual reality (VR) simulators, for instance, enable trainees to practice procedures in a risk-free, computer-generated environment [12,13,14,15]. VR training platforms can offer unlimited repetitions, objective performance metrics, and immediate feedback for certain skills [16]. However, because they lack any genuine haptic feedback, purely virtual systems cannot replicate the tactile experience and complex instrument handling of real orthopedic surgery [8, 16]. Augmented reality (AR) systems, on the other hand, overlay virtual information onto the real operative field and have primarily been used as navigational aids in surgery [17]. By themselves, AR applications do not provide a fully immersive hands-on practice environment or real-time performance coaching for the trainee. As a result, none of the existing simulation modalities fully integrates realistic tactile practice with interactive guidance, leaving a significant gap in current surgical training tools.

Immersive mixed reality (IMR) has emerged as a promising approach to bridge this gap in training [1, 18]. By wearing an IMR head-mounted display, a trainee can see and interact with 3D holographic content superimposed onto real-world objects in the training space. This technology uses spatial mapping to anchor virtual objects to physical counterparts, allowing, for instance, a trainee to practice on a tangible anatomical phantom while receiving holographic guidance or augmented feedback aligned in real time. IMR has already shown value in the operating room as a navigation aid, with studies validating its accuracy and feasibility for guided surgeries [4, 13, 15]. However, its effectiveness in structured surgical education and skill acquisition has yet to be established. In fact, a recent review noted that only 22.7% of studies involving XR technologies have focused on practical skills training applications [14]. When combined with a physical phantom, IMR offers the possibility of an immersive training environment that also preserves authentic haptic feedback for higher fidelity practice [13]. Nonetheless, most reported IMR-based training systems to date have primarily focused on static navigational overlays rather than providing interactive instructional feedback based on the trainee’s real-time performance [8, 19] leading to inadequate integration between virtual simulation and physical hands-on training.

Therefore, this study proposes a novel IMR self-training system (IMR-SS) for pedicle screw placement. This system provides a fully integrated training environment that bridges the virtual and physical worlds. The IMR-SS combines the trainee’s hands-on performance with a dynamic holographic virtual environment via visual tracking, which monitors the spatial movements of surgical instruments and the physical phantom to deliver real-time guidance and navigation cues. To our knowledge, this IMR-SS is the first immersive mixed-reality self-training platform for pedicle screw insertion, and its effectiveness has been evaluated in a randomized controlled trial.

Methods

We developed an immersive mixed-reality system for pedicle-screw training (IMR-SS) comprising three modules—Teaching, Hands-on, and Assessment—and evaluated its efficacy in a randomized controlled trial. In evaluation, thirty-two undergraduate medical students from two medical schools participated to compare the IMR-SS with a traditional method in terms of training outcomes and satisfaction.

Development of IMR-SS

Guided by the Cognitive Affective Model of Immersive Learning [20] we designed an immersive mixed-reality system for pedicle-screw training (IMR-SS) with three modules—Teaching, Hands-on, and Assessment (Fig. 1) to ensure that guidance and metrics remained consistent across virtual and physical contexts. A browser-based demo is provided in the Supplementary Material.

Teaching Module: An expert-validated curriculum was implemented on HoloLens 2 using Unity 2021.3.9 LTS and MRTK 2.8.2, enabling seamless virtual–physical overlap with multimodal interaction (voice, gesture, eye tracking) [20, 21]. Training proceeds through four stages—Preparation, Exploration, Exercise, Reflection. Learners (i) review instruments & anatomy, (ii) study nine procedural steps for pedicle screw insertion, (iii) operate on a tracked phantom with real-time holographic guidance, and (iv) analyze automatically generated logs and reports (Supplementary Material, Training Materials).

Hands-on Module: A modular lumbar phantom (L1–L5) was reconstructed from patient CT data [22]. Replaceable PLA vertebrae reproduce human bone mechanics [23], while muscle and vascular components provide landmarks. Custom fiducials on each instrument are tracked by a binocular camera, feeding 6-DoF poses to a digital-twin replica and keeping tactile practice synchronized with virtual cues. Trackers were mounted on real surgical instruments to enable real-time, precise spatial tracking during tactile practice.

Assessment Module: Evaluation metrics for pedicle screw placement were defined with expert orthopedic surgeons, and these thresholds also informed the real-time instruction method implemented in the Teaching Module. Real-time metrics quantify precision (pose deviation from expert-defined thresholds) and safety (instrument–tissue collisions). Instrument poses are mapped into the digital twin, and deviations immediately trigger on-device warnings and adaptive guidance (Supplementary Material, Training Materials, Pedicle Screwing Procedure), closing the feedback loop across all three modules.

[IMAGE OMITTED: SEE PDF]

The arrows indicate the implementation order and the flow of data. The 3 modules are shown in different colors: the hands-on module in red, the assessment module in yellow, and the teaching module in green.

Study design

The IMR-SS system was evaluated with ethics approval from Peking Union Medical College Hospital (K5533-K24C0630) and The First Affiliated Hospital of Xi’an Jiaotong University (XJTU1AF2024LSYY-097). Thirty-two undergraduate clinical medical students who had completed systematic anatomy and had no prior learning/observation experience with this surgery participated. The trial used a parallel design with random allocation to the IMR group or the control group (16 per group; 8 from each center). Randomization was performed by an independent researcher using a computer-generated sequence. The study ran from 3 to 28 January 2024 in classroom settings. Written informed consent was obtained from all participants. Outcome assessors were blinded. Table 1 presents baseline characteristics.

[IMAGE OMITTED: SEE PDF]

The experiment comprised four sequential sessions (Figs. 2 and 3): an interaction learning session to familiarize participants with the interface; a learning session on pedicle screw placement; an exercise session on a physical phantom (PP); and an assessment session with knowledge and skills evaluation. All devices and equipment were fully tested before the experiment. Session timing was controlled by the conductor without intervention.

In the interaction learning session, both groups first used a content-free (“blank”) IMR system that preserved the IMR-SS structure but contained no curricular content, ensuring an identical cognitive load before surgical training.

In the learning session, both groups learned pedicle screw placement in three timed sections: instrument preview (8 min), model preview (5 min), and procedure instructions (15 min). The IMR group used the IMR-SS; the control group used a digital textbook arranged with the same content and section timing. Once a time limit was reached, participants moved to the next section.

In the exercise session, participants inserted one screw into the L3 vertebra on the PP for 15 min. The IMR group practiced with the IMR-SS (Fig. 3); the control group practiced with reference to an operation animation clip (Fig. 3).

In the assessment session, participants first completed a 12-item multiple-choice quiz (3 anatomy; 9 procedures/instruments; Maximum 10 min; Supplementary Material, Postlearning Quiz). They then placed a new screw on L3 independently with no time limit, following the tutorial content and techniques.

After the assessment, both groups completed Likert-scale questionnaires on (i) perceived confidence in mastering pedicle screw placement, (ii) satisfaction with the training experience, and (iii) system usability (NASA-TLX workload).

[IMAGE OMITTED: SEE PDF]

Data analysis

According to the Assessment Module, the pedicle-screw procedure was decomposed into sequential nodes, each ending in a benchmark event (e.g., “screw fully seated in the pedicle”). For every node, we prospectively listed mandatory actions—instrument selection, assembly, and operative gesture—required for full credit.

During the exercise session, the IMR system logged 6-DoF instrument poses via visual tracking and auto-flagged any pedicle breach or collision with protected tissue; these were counted as safety errors. In the assessment session trainees operated unaided (without HoloLens 2) to test transfer. Trial videos were independently reviewed by two blinded spine surgeons; any omitted or mis-used step was recorded as a procedural error, and disagreements were resolved by consensus. Total errors equaled safety and procedural errors. For completion rate, node-level completion was:

$$\:CR_{substep}=\left\{\begin{array}{c}\frac{Number\:of\:Completed\:Node\:Tasks}{Number\:of\:Node\:Tasks}\;Benmark\:Achievced\\\:Benmark\:Not\:Achievced)\end{array}\right.$$

Overall completion for a trial was the arithmetic mean across nodes.

$$\:CR=\frac{{CR}_{\text{s}\text{u}\text{b}\text{s}\text{t}\text{e}\text{p}1}+{CR}_{\text{s}\text{u}\text{b}\text{s}\text{t}\text{e}\text{p}2}+\dots\:+{CR}_{substepn}}{n}$$

CT scans of the L3 models with screws in situ were obtained after the exercise session and assessment session. Screw orientation was measured by the traditional method [24] as the angle between the vertebral reference axis and the screw on the axial (transverse) and sagittal (lateral) planes (transverse and lateral angles). Prior CT-based studies show that larger deviations increase the risk of pedicle-wall breach and adjacent-segment/facet injury [25, 26]; therefore we analyzed the absolute angle values as the placement-quality metric, with smaller absolute angles indicating better insertion.

An a priori power analysis (two-tailed, α = 0.05, β = 0.10, power = 90%) indicated that 15 participants per group were required to detect an absolute accuracy difference of 5% (pooled SD = 0.15, effect size δ = 0.32). The sample size was calculated by:

$$\:N=\frac{2{\sigma\:}^{2}f(\alpha\:,\beta\:)}{{\delta\:}^{2}},\:\:where\:f(\alpha\:,\beta\:)\:=\:10.5.$$

Comparisons used Fisher’s exact, independent two-tailed t tests or Mann–Whitney U tests as appropriate; significance \(\:p<0.05\). Analyzes were performed in SPSS, Excel, and R.

[IMAGE OMITTED: SEE PDF]

Results

All participants in the experiment were medical students with no prior orthopedic surgical training experience and completed all the sessions without any expert guidance. During the learning session, both the IMR group and the control group were provided with identical learning materials and followed the same sequence of content. The duration allocated to each learning section was standardized across both groups. To maintain consistency and prevent any distractions during the learning process, no data were recorded or collected during this phase. However, post-experiment analysis included evaluating the participants’ satisfaction during the learning session. Our analysis focused on trainees’ performance during the exercise session and assessment session.

Performance during training and assessment sessions

In the exercise session, there was no significant difference between the two groups in terms of placement angle. However, 25% (4/16) of the participants in the control group experienced breaches, whereas none of the participants in the IMR group experienced breaches (p = 0.051). During the assessment session, 12.5% (2/16) of the participants in the control group breached, whereas none of the participants in the IMR group breached. In the assessment session, the transverse angle in the IMR group (13.88°±6.98°) was significantly smaller than that in the control group (20.89°±11.59°, p = 0.049). Additionally, the variance in the transverse and lateral screw placement angles was significantly smaller in the IMR group than in the control group (Table 2), indicating greater performance stability in the IMR group than in the control group (Fig. 5).

During the assessment session, the completion rate (median (IQR)) in the IMR group was 1.000 (1.0–1.0), significantly higher than 0.88 [0.80,1.00] in the control group (p < 0.01), as shown in Fig. 4. The median number of errors per participant was 0 [0, 0] in the IMR group versus 2 [1, 3.8] in the control group (p < 0.01). Most errors in both groups occurred in the last step, which involved more instruments and node tasks. Errors were categorized into instrument-usage errors, procedure-benchmark errors, and node-task errors; the IMR group committed significantly fewer errors in all categories (details in Table 2).

[IMAGE OMITTED: SEE PDF]

Box-and-whisker plots with overlaid scatter show the distribution of procedural error count and completion rate (CR) during the assessment session for the control group (n = 16) and the IMR group (n = 16). Each dot represents one participant. The boxes indicate the median and inter-quartile range (IQR); whiskers extend to data within 1.5 × IQR. Because the variables were not normally distributed, between-group differences were evaluated with the two-tailed Mann–Whitney U test. Exact p-values are reported in Table 2.

Additionally, for the IMR group, the transverse angles in the exercise session were significantly smaller than those in the assessment session (11.07°±6.88° vs. 13.88°±6.98°, p = 0.040; Fig. 5). We assumed that the real-time instructions and navigation provided by the IMR-SS helped the IMR group achieve better placement results during the exercise session. Despite the decrease, the transverse angle in the IMR group remained significantly smaller than that in the control group (13.88°±6.98° vs. 20.89°±11.59°, p = 0.049) during the assessment session.

[IMAGE OMITTED: SEE PDF]

Box-and-whisker plots with overlaid scatter compare screw placement angle errors for the control group (n = 16) and the IMR group (n = 16) across Exercise and Assessment sessions. The left panel depicts errors in the Transverse plane; the right panel depicts errors in the Lateral plane. If the data met normality assumptions, a two-tailed independent-samples t-test was applied. See Table 2 for exact p-values. Plot elements (median, IQR, whiskers, scatter) follow the same conventions as in Fig. 2.

There was no significant difference between the two groups in the quiz session. The participants might have relevant knowledge as they had learned systematic anatomy (Supplementary Material, Table 1). Additionally, the results revealed that the IMR-SS system might not offer significant advantages over traditional methods in terms of general knowledge learning [20].

[IMAGE OMITTED: SEE PDF]

Training experience and satisfaction

The training experience of the IMR-SS was evaluated through subjective Likert-scale questionnaires. We collected the perceived scores of confidence in the mastering pedicle screw placement, the satisfaction score of the training experience, and the usability score of the system. For perceived confidence, the perceived score of the IMR group was significantly greater than that of the control group (instrument usage: 4.81 ± 0.40 vs. 4.19 ± 0.66, p = 0.003; autonomy: 4.56 ± 0.51 vs. 4.00 ± 0.89, p = 0.037; procedure: 4.81 ± 0.40 vs. 4.31 ± 0.60, p = 0.010). For training satisfaction, the perceived scores of curriculum satisfaction in the IMR Group were significantly higher than those in the control group (“I believe the teaching content is accurate”: 4.69 ± 0.48 vs. 4.19 ± 0.54, p = 0.010; “I believe the teaching content is very helpful”: 4.38 ± 0.62 vs. 4.81 ± 0.54, p = 0.042; “I believe the teaching content is very practical”: 4.81 ± 0.54 vs. 4.38 ± 0.62 , p = 0.042). The IMR group also reported significantly greater useability (total score: 42.44 ± 4.50 vs. 38.75 ± 5.53, p = 0.047). The score of each item in the questionnaire can be found in the Supplementary Material, Table 4. Notably, the IMR group reported greater confidence in using the IMR-SS than in the digital textbook (“I feel very confident using this tutorial”: 4.63 ± 0.50 vs. 3.56 ± 0.89, p < 0.01). A detailed description of the data can be found in the Supplementary Material, Table 4.

Discussion

In our randomized trial, trainees in both groups first completed an interactive orientation session to equalize baseline cognitive load. They then underwent the same duration of pedicle screw training, allowing a direct comparison of performance during the subsequent exercise and assessment sessions. The IMR-SS curriculum consisted of three modules—teaching, hands-on practice, and assessment—seamlessly integrating virtual guidance with physical practice. Our findings demonstrate that the IMR-SS significantly improved both training effectiveness and the learning experience compared to traditional materials, providing evidence to support incorporating IMR technologies into surgical education programs.

Training outcomes: performance, accuracy, knowledge, and satisfaction

In terms of procedural skill acquisition, the IMR group made significantly fewer errors and achieved a higher task completion rate than the control group. The IMR-SS provided multi-perspective views of each step with synchronized visual, text, and voice instructions. According to multimodal learning theory [21] presenting information through multiple sensory channels enhances cognitive processing and promotes more effective deliberate practice. Unlike conventional VR or screen-based simulators for pedicle screwing [13, 27] our IMR-SS integrates a patient-specific physical phantom, ensuring true-to-scale haptic feedback in a real-world setting. Prior studies note that purely screen-based or virtual systems can cause proprioceptive dissonance and spatial perception mismatches for trainees [28]. By allowing trainees to use actual instruments on a tangible model while receiving holographic feedback, IMR-SS avoids such issues. This high-fidelity tactile interaction may enhance the realism of training and improve skill transfer to the operating room, consistent with evidence that adding haptic feedback in simulation improves technical performance outcomes [29]. A recent randomized laboratory study showed that VR pedicle-screw simulation significantly improved placement accuracy and reduced cortical breaches versus sawbone practice [30] and a multicenter RCT reported higher excellent/good accuracy with AR-guided fixation than freehand (99.1% vs. 91.7%) [31], consistent with our IMR group’s lower errors and higher completion.

During the unassisted assessment session, trainees who used IMR-SS achieved better pedicle screw trajectories, including a more optimal transverse angle and fewer cortical breaches, than the control group. This improvement in spatial accuracy after short-term training is similar to previous reports in which virtual reality (VR) training enhanced learners’ visuospatial skills and precision in orthopedic procedures [12, 32, 33]. For example, immersive simulations have been shown to improve the angle accuracy of screw insertion and spatial awareness during surgical tasks [34]. Although some studies have cautioned that excessive navigational cues in AR/MR systems may lead to over-reliance on guidance [15, 19] our results indicate that the IMR group retained significantly better freehand accuracy than the control group even when neither wore the headset. Providing precise visual–spatial cues in early training can help novices form accurate mental models of the task, in line with cognitive load theory [35] As training progresses, a deliberate reduction of visual aids through a hierarchical instructional design allows skills to gradually transfer from the virtual context to real-world application [36, 37]. This phased approach likely prevented over-dependence on the holographic cues and enabled trainees to perform with confidence without assistance.

There was no significant difference between the IMR and control groups in the post-training theoretical knowledge quiz. This aligns with existing research suggesting that XR technologies do not necessarily outperform traditional study methods for factual knowledge such as anatomy or terminology [20, 38]. In our study, both groups received the same textual and graphic content, just delivered through different mediums, which explains the equivalent scores. Indeed, while IMR provides rich 3D visuals, factual knowledge in our module was still conveyed via text and images, and overwhelming trainees with holographic detail might even hinder memorization of facts. Nonetheless, the IMR group reported greater satisfaction with the training and higher confidence in their mastering pedicle screwing technique. This subjective improvement is consistent with previous immersive VR studies showing that a heightened sense of presence and interactivity can increase learners’ engagement and self-confidence even if test scores remain similar [20]. In our case, the IMR-SS’s immersive teaching likely made the learning process more enjoyable and motivating for the students. Future refinements of the system could explore improved visual representations and optimized multimodal interactions to minimize cognitive overload, thereby better balancing the delivery of procedural skills versus didactic knowledge.

Educational and practical benefits of the modular IMR-SS training framework

Our results demonstrate that many principles from VR-based surgical training can be effectively translated into an immersive mixed reality context. By fostering a closer connection to the real world, IMR allows trainees to acquire skills in an authentic environment and enhances the transfer of those skills to clinical practice. This supports previous suggestions that increasing the physical realism of simulation (e.g., by using real instruments and patient-specific models) improves skill transfer to the operating room [1, 13]. At the same time, prior studies of augmented and mixed reality training systems have reported mixed results, which researchers attribute in part to a lack of comprehensive, systematic training frameworks [12, 13, 39]. We addressed this gap by introducing a modular IMR-SS curriculum that tightly integrates teaching, hands-on, and assessment phases, effectively bridging the gap between virtual simulation and physical practice. This structured approach provides a consistent learning trajectory and may be one reason why our IMR-SS yielded significant performance benefits where some earlier, less structured MR training attempts did not. A well-designed, theory-driven curriculum appears crucial for achieving reliable training outcomes with XR technologies. We believe that our integrated framework can serve as a model for developing more ubiquitous and effective IMR-based training programs.

Each module of the IMR-SS was designed with proven educational strategies in mind. In the teaching module, we implemented a cycle of Preparation – Exploration – Exercise – Reflection to engage trainees in experiential learning [40]. The pedicle screw procedure was broken down into discrete tasks after an expert-led cognitive task analysis [41]. This stepwise decomposition of the surgery not only made the process less overwhelming for novices but also ensured that critical sub-tasks received adequate attention and practice [42]. The hands-on module leveraged patient-specific 3D-printed lumbar models paired with their holographic digital twins, enabling seamless integration of the virtual and physical worlds. Using spatial tracking cameras, we brought real surgical tools into the simulation environment, so trainees could perform the procedure on a tangible spine while seeing aligned holographic guidance. This design provides authentic tactile feedback absent in purely virtual trainers and helps bridge the “reality gap” that often exists in VR simulators. By practicing with real instruments on anatomically accurate models, users may more readily transfer their motor skills to actual patients. In the assessment module, we evaluated adherence to standard procedure and technique quality through an accuracy-and-safety analysis. Unlike many prior simulator studies that looked only at final outcomes (e.g. screw placement accuracy), our assessment also tracked any errors or breaches made during the process [15, 19, 39]. Moreover, the IMR-SS provided concurrent real-time multimodal feedback (visual warnings, auditory prompts, and haptic cues) during training, allowing participants to immediately recognize and correct mistakes. Such immediate feedback has been shown to be especially beneficial for novices, accelerating skill acquisition and preventing the reinforcement of bad habits [43]. Taken together, these features of our framework (task deconstruction, physical–virtual integration, and continuous feedback) differentiate the IMR-SS from earlier MR training setups and likely contributed to its training efficacy.

In addition to educational benefits, the IMR-SS offers practical and economic advantages compared to many existing simulation options. The system runs on a desk-sized platform with a standard PC and a consumer-grade mixed reality headset, plus two tracking cameras. The main hardware investment (approximately US$4,500 total) is moderate relative to high-end surgical simulators or fully immersive VR labs. Meanwhile, the recurring cost per use is kept low by using inexpensive 3D-printed vertebral models (roughly US$0.50 in materials per case) in place of costly cadaveric specimens or elaborate plastic replicas. This allows for unlimited practice and repeatability on patient-specific anatomies without significant additional expense. While purely software-based VR trainers have even lower marginal costs, they sacrifice the haptic realism that IMR-SS provides. Our approach thus attempts to balance fidelity with affordability. Future cost–benefit analyses are warranted to determine whether the improved training outcomes and reusable components of IMR-SS translate into overall cost effectiveness for surgical education programs. For widespread adoption, institutions will need to assess if the up-front investment in MR hardware can be justified by the reduction in faculty supervision time, cadaver use, or trainee errors achieved through self-training on platforms like IMR-SS.

Limitations

Our study has several limitations. We did not include transfer studies on cadavers or in the operating room. However, the use of physical phantoms derived from real cases and authentic surgical instruments in the IMR-SS could enhance the transferability of acquired skills [44]. Our study revealed significantly better pedicle screw placement in short-term training with novice undergraduate students using the IMR-SS, but further research should investigate whether the IMR-SS can foster a shorter learning curve. Our study did not evaluate soft tissue handling or non-technical skills, such as risk decision-making and teamwork. As they are fundamental skills in surgical practice [4, 45] future research could incorporate emergency event simulations and multiplayer cooperation into the IMR-SS to assess their effectiveness in enhancing these nontechnical skills.

Our study investigated the effectiveness of the IMR-SS for pedicle screwing, a classic procedure in spine surgery [46]. This serves as a successful case for designing IMR-SSs for more complex surgical training. Future research will develop IMR-SSs for more specific and complex open surgeries and explore their effectiveness in a wider range of surgical procedures to validate their generalizability.

Conclusions

This randomized controlled trial demonstrated that the Immersive Mixed Reality Surgical Self-Training System (IMR-SS) significantly enhances the precision and safety of pedicle screw placement among novice medical students. The IMR-SS effectively integrates virtual instructions with hands-on practice to improve surgical training outcomes and provides a modularized framework to build an IMR surgical self-training system. Future research should explore long-term skill retention, the applicability of the IMR-SS to a wider range of surgical procedures, and the scalability and cost-effectiveness of broader medical education.

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

References

Wang Z, Shen J. Simulation training in spine surgery. J Am Acad Orthop Surg. 2022;30(9):400–8. https://doi.org/10.5435/JAAOS-D-21-00756.

LaPorte DM, Tornetta P, Marsh JL. Challenges to orthopaedic resident education. J Am Acad Orthop Surg. 2019;27(12):419–25. https://doi.org/10.5435/JAAOS-D-18-00084.

Bashankaev B, Baido S, Wexner SD. Review of available methods of simulation training to facilitate surgical education. Surg Endosc. 2011;25(1):28–35. https://doi.org/10.1007/s00464-010-1123-x.

Viglialoro RM, Condino S, Turini G, Carbone M, Ferrari V, Gesi M. Augmented reality, mixed reality, and hybrid approach in healthcare simulation: a systematic review. Appl Sci. 2021;11(5):2338. https://doi.org/10.3390/app11052338.

Gonzalvo A, Fitt G, Liew S, et al. The learning curve of pedicle screw placement: how many screws are enough?? Spine. 2009;34(21):E761–5. https://doi.org/10.1097/BRS.0b013e3181b2f928.

Karam MD, Pedowitz RA, Natividad H, Murray J, Marsh JL. Current and future use of surgical skills training laboratories in orthopaedic resident education: a national survey. J Bone Joint Surg. 2013;95(1):e4. https://doi.org/10.2106/JBJS.L.00177.

Condino S, Turini G, Mamone V, Parchi PD, Ferrari V. Hybrid spine simulator prototype for X-ray free pedicle screws fixation training. Appl Sci. 2021;11(3):1038. https://doi.org/10.3390/app11031038.

Condino S, Turini G, Parchi PD, et al. How to build a patient-specific hybrid simulator for orthopaedic open surgery: benefits and limits of mixed-reality using the Microsoft HoloLens. J Healthc Eng. 2018;2018:1–12. https://doi.org/10.1155/2018/5435097.

Tack P, Victor J, Gemmel P, Annemans L. 3D-printing techniques in a medical setting: a systematic literature review. BioMed Eng OnLine. 2016;15(1):115. https://doi.org/10.1186/s12938-016-0236-4.

Rosenbaum L. Leaping without looking–duty hours, autonomy, and the risks of research and practice. N Engl J Med. 2016;374(8):701–3. https://doi.org/10.1056/NEJMp1600233.

Sadati L, Yazdani S, Heidarpoor P. Exploring the surgical residents’ experience of teaching and learning process in the operating room: a grounded theory study. J Educ Health Promot. 2021;10(1):176. https://doi.org/10.4103/jehp.jehp_1007_20.

McCloskey K, Turlip R, Ahmad HS, Ghenbot YG, Chauhan D, Yoon JW. Virtual and augmented reality in spine surgery: a systematic review. World Neurosurg. 2023;173:96–107. https://doi.org/10.1016/j.wneu.2023.02.068.

Nazzal EM, Zsidai B, Hiemstra LA, Lustig S, Samuelsson K, Musahl V. Applications of extended reality in orthopaedic surgery. J Bone Joint Surg. 2023;105(21):1721–9. https://doi.org/10.2106/JBJS.22.00805.

Tang YM, Chau KY, Kwok APK, Zhu T, Ma X. A systematic review of immersive technology applications for medical practice and education - Trends, application areas, recipients, teaching contents, evaluation methods, and performance. Educ Res Rev. 2022;35:100429. https://doi.org/10.1016/j.edurev.2021.100429.

Azad TD, Warman A, Tracz JA, Hughes LP, Judy BF, Witham TF. Augmented reality in spine surgery – past, present, and future. Spine J. 2024;24(1):1–13. https://doi.org/10.1016/j.spinee.2023.08.015.

Imperatori A, Grossi S, Cattoni M, Rotolo N. The role of haptic feedback in video-assisted thoracic surgery simulation training. Shanghai Chest. 2022;6:11–11. https://doi.org/10.21037/shc-22-18.

Asoodar M, Janesarvatan F, Yu H, De Jong N. Theoretical foundations and implications of augmented reality, virtual reality, and mixed reality for immersive learning in health professions education. Adv Simul. 2024. https://doi.org/10.1186/s41077-024-00311-5.

Anderson MJJ, deMeireles AJ, Trofa DP, Kovacevic D, Ahmad CS, Lynch TS. Cognitive training in orthopaedic surgery. JAAOS Glob Res Rev. 2021;5(3). https://doi.org/10.5435/JAAOSGlobal-D-21-00021.

Applications of augmented. And virtual reality in spine surgery and education: A review. Seminars Spine Surg. 2021;33(2):100875. https://doi.org/10.1016/j.semss.2021.100875.

Makransky G, Petersen GB. The cognitive affective model of immersive learning (CAMIL): a theoretical research-based model of learning in immersive virtual reality. Educ Psychol Rev. 2021;33(3):937–58. https://doi.org/10.1007/s10648-020-09586-2.

Di Mitri D, Schneider J, Limbu B, Mat Sanusi KA, Klemke R. Multimodal learning experience for deliberate practice. In: Giannakos M, Spikol D, Di Mitri D, Sharma K, Ochoa X, Hammad R, editors. The multimodal learning analytics handbook. Springer International Publishing; 2022. pp. 183–204. https://doi.org/10.1007/978-3-031-08076-0_8.

Li J, An Z, Wu J, et al. Construction of the adjusted scoliosis 3D finite element model and biomechanical analysis under gravity. Orthop Surg. 2023;15(2):606–16. https://doi.org/10.1111/os.13572.

Bohl MA, Morgan CD, Mooney MA, et al. Biomechanical testing of a 3D-printed L5 vertebral body model. Cureus. 2019;11(1). https://doi.org/10.7759/cureus.3893.

Kim MC, Chung HT, Cho JL, Kim DJ, Chung NS. Factors affecting the accurate placement of percutaneous pedicle screws during minimally invasive transforaminal lumbar interbody fusion. Eur Spine J. 2011;20(10):1635–43. https://doi.org/10.1007/s00586-011-1892-5.

Sterba W, Kim DG, Fyhrie DP, Yeni YN, Vaidya R. Biomechanical analysis of differing pedicle screw insertion angles. Clin Biomech. 2007;22(4):385–91. https://doi.org/10.1016/j.clinbiomech.2006.11.007.

Wang Q, Gao Z, Guo K, Wang F, Wu D. Effect of sagittal screw angle and distance of screw apex to superior endplate on adjacent segment disease after posterolateral lumbar fusion: a retrospective study. J Orthop Surg Res. 2022. https://doi.org/10.1186/s13018-022-03383-z.

Gardeck AM, Pu X, Yang Q, Polly DW, Jones KE. The effect of simulation training on resident proficiency in thoracolumbar pedicle screw placement using computer-assisted navigation. J Neurosurgery: Spine. 2021;34(1):127–34. https://doi.org/10.3171/2020.5.SPINE2067.

Lohmann J, Butz MV. Lost in space: multisensory conflict yields adaptation in spatial representations across frames of reference. Cogn Process. 2017;18(3):211–28. https://doi.org/10.1007/s10339-017-0798-5.

Impact of haptic feedback on surgical training outcomes. A randomised controlled trial of haptic versus non-haptic immersive virtual reality training. Annals Med Surg. 2022;83:104734. https://doi.org/10.1016/j.amsu.2022.104734.

Beyond Traditional Training. Exploring the benefits of virtual reality simulator in lumbar pedicle screw Insertion - A randomized controlled trial. World Neurosurg. 2024;189:e61–8. https://doi.org/10.1016/j.wneu.2024.05.163.

Ma Y, Wu J, Dong Y, Tang H, Ma X. Augmented reality navigation system enhances the accuracy of spinal surgery pedicle screw placement: a randomized, multicenter, parallel-controlled clinical trial. Orthop Surg. 2025;17(2):631–43. https://doi.org/10.1111/os.14295.

Cevallos N, Zukotynski B, Greig D, Silva M, Thompson RM. The utility of virtual reality in orthopedic surgical training. J Surg Educ. 2022;79(6):1516–25. https://doi.org/10.1016/j.jsurg.2022.06.007.

Logishetty K, Gofton WT, Rudran B, Beaulé PE, Cobb JP. Fully immersive virtual reality for total hip arthroplasty: objective measurement of skills and transfer of visuospatial performance after a competency-based simulation curriculum. J Bone Joint Surg. 2020;102(6):e27. https://doi.org/10.2106/JBJS.19.00629.

Xin B, Chen G, Wang Y, et al. The efficacy of immersive virtual reality surgical simulator training for pedicle screw placement: A randomized Double-Blind controlled trial. World Neurosurg. 2019;124:e324–30. https://doi.org/10.1016/j.wneu.2018.12.090.

Anderson MJJ, deMeireles AJ, Trofa DP, Kovacevic D, Ahmad CS, Lynch TS. Cognitive training in orthopaedic surgery. JAAOS: Global Research and Reviews. 2021;5(3):e2100021. https://doi.org/10.5435/JAAOSGlobal-D-21-00021.

So HY, Chen PP, Wong GKC, Chan TTN. Simulation in medical education. J R Coll Physicians Edinb. 2019;49(1):52–7. https://doi.org/10.4997/jrcpe.2019.112.

Briggs LJ. Instructional design: principles and applications. Educational Technology; 1991.

Makransky G, Andreasen NK, Baceviciute S, Mayer RE. Immersive virtual reality increases liking but not learning with a science simulation and generative learning strategies promote learning in immersive virtual reality. J Educ Psychol. 2021;113(4):719–35. https://doi.org/10.1037/edu0000473.

Farah GJ, Rogers JL, Lopez AM, et al. Resident training in spine surgery: a systematic review of simulation-based educational models. World Neurosurg. 2023;174:81–115. https://doi.org/10.1016/j.wneu.2023.03.032.

Morris TH. Experiential learning – a systematic review and revision of Kolb’s model. Interactive Learning Environments. Published online November 16, 2020. Accessed December 19, 2024. https://www.tandfonline.com/doi/abs/https://doi.org/10.1080/10494820.2019.1570279

Utilization of a. Cognitive task analysis for laparoscopic appendectomy to identify differentiated intraoperative teaching objectives. Am J Surg. 2012;203(4):540–5. https://doi.org/10.1016/j.amjsurg.2011.11.002.

Paas FGWC, Van Merriënboer JJG. Instructional control of cognitive load in the training of complex cognitive tasks. Educ Psychol Rev. 1994;6(4):351–71. https://doi.org/10.1007/BF02213420.

Sigrist R, Rauter G, Riener R, Wolf P. Augmented visual, auditory, haptic, and multimodal feedback in motor learning: a review. Psychon Bull Rev. 2013;20(1):21–53. https://doi.org/10.3758/s13423-012-0333-8.

Bohl MA, McBryan S, Pais D, et al. The living spine model: a biomimetic surgical training and education tool. Oper Neurosurg. 2020;19(1):98–106. https://doi.org/10.1093/ons/opz326.

Pakkasjärvi N, Anttila H, Pyhältö K. What are the learning objectives in surgical training – a systematic literature review of the surgical competence framework. BMC Med Educ. 2024;24(1):119. https://doi.org/10.1186/s12909-024-05068-z.

Park SM, Shen F, Kim HJ, et al. How many screws are necessary to be considered an experienced surgeon for freehand placement of thoracolumbar pedicle screws? Analysis using the cumulative summation test for learning curve. World Neurosurg. 2018;118:e550–6. https://doi.org/10.1016/j.wneu.2018.06.236.