During the initial preparation process of this study, the orthopedic bone and soft tissue tumor team at Peking Union Medical College Hospital provided relevant data. The medical team exemplified surgical instruments used for spinal tumors and percutaneous vertebroplasty, elucidated the surgical steps, and demonstrated the operative process. Based on this premise, the team from the Fundamental Industry Training Center of Tsinghua University utilized simulation technique to create interactive anatomical program of spinal tumor and surgical operation procedure for percutaneous vertebroplasty (PVP). This primarily involved script development for AR-assisted spinal tumor and percutaneous vertebroplasty surgery, calibrating the position of anatomical structures, measuring and reconstructing surgical instruments, and establishing multi-person collaboration and position sharing (Figure 1). Construction of spinal tumor model: Firstly, with informed consent from the patient and approval from the hospital’s ethics committee, Digital Imaging and Communications in Medicine (DICOM) files were obtained. Since the DICOM files were 2D image files, the creation of the three-dimensional mesh was performed using “3D Slicer” software, which can be considered as reverse engineering. Subsequently, the model was refined and reconstructed using “3Ds Max” software; Secondly, we have taken the following steps to optimize the mesh: (1) Retopology: The mesh topology was optimized; (2) Manual simplification: Unnecessary meshes were removed, and structures with poor display quality were optimized; (3) Structural adjustment: Manually drawn models can have deviations, similar to mechanical assembly errors. Using commands like lattice deformation, details were smoothly adjusted to ensure complete integration of the lumbar vertebrae; (4) Boolean operations: Overlapping parts of two models were removed. For example, independently drawn spinal tumor and vertebral body are separate parts and will overlap when combined. Using Boolean algorithms, appropriate spaces were cut from the vertebral body to accommodate the spinal tumor. Subsequently, we used Unity program to create the AR environment and implement the communication functionality through code, then finally deployed it to HoloLens 2 for each student to experience after multiple debugging. We reconstructed surgical instruments for percutaneous vertebroplasty surgery, includeing drapes, anesthesia syringe, needle, puncture cannula, drill, sampler, bone cement, push rod, gauze, etc. The tasks of the teaching groups were to help students master the anatomical structure of the spine, spinal tumor, and learn the PVP surgical procedure. The above models all have natural interaction functions such as finger dragging, zooming, and tapping, for students to simulate and operate. In addition, the program has a visual projection function, which allows for viewing of spinal tumor lesions and surrounding structures under the guidance of teachers, as well as demonstrating the surgical process. The perspective mode is akin to viewing a real human body, where internal organs and bone structures are not visible beneath the skin. When inserting medical instruments into the human model, students rely on experience and medical imaging references. However, the perspective mode is different. It assists students by providing a three-dimensional perspective view, allowing them to clearly and intuitively see through the skin to observe the vertebral structure and spinal tumor. Through close collaboration, researchers presented spinal tumor and percutaneous vertebroplasty by using augmented reality (Figures 2, 3).

To further confirm the application effect and feasibility of AR technology in teaching spinal tumor’s anatomy and surgery, 8-year program medical students were selected from Peking Union Medical College and Tsinghua University who met the inclusion criteria in the 8-year clinical medicine program as research subjects. The inclusion criteria for this study are as follows: students who have studied for a minimum of 5 years, and have not previously undergone any spinal tumor surgery or virtual reality/augmented reality-assisted surgery. Exclusion criteria: students who could not complete the 2-h teaching plan on time, could not participate in post-teaching evaluations, were in a grade of fewer than 5 years, had previously participated in spinal tumor surgery, or received surgical simulation training.

The teacher employs a uniform teaching curriculum to instruct involved students on fundamental theories and anatomical structures of spinal tumor. This encompasses topics such as spinal anatomy, spinal tumor anatomy, prevalent disorders associated with spinal tumors, spinal tumor metastasis, standard surgical procedures, surgical indications and contraindications, details for percutaneous vertebroplasty, and other specialized knowledge. Following the completion of the instruction, a group simulation lesson will be carried out for all students in two groups.

The students included in this study will be randomly divided into a traditional teaching group (Group A, using traditional teaching methods in the simulation lesson) and a AR-assisted teaching group (Group B, using head-worn HoloLens 2 holographic glasses for AR teaching in the simulation lesson). The same senior physician will teach the two groups of students, and the teaching staff will have the same class hours, courseware, basic knowledge explanation, and model teaching aids application, except for AR-related equipment. The teacher in the traditional teaching group (Group A) explained all same procedures using 3D-printed lumbar spine anatomy models, PVP surgical tools, and solid human anatomy models. The students in group B experienced the spinal tumor instance model and PVP surgical operation process using augmented reality technology and head-worn HoloLens holographic glasses. As they go through a spinal tumor instance model, students can change the view from perspective to non-perspective, remove, enlarge, and decrease the vertebrae and attachments of each segment. The layout and structure of the spinal trabeculae, as well as the location and invasive features of spinal tumors, can be intuitively felt by students when they switch to perspective mode. Based on their learning objectives, students can control their learning process, rewatch crucial steps, and switch to non-perspective or perspective modes during the PVP surgical simulation demonstration.

When the regular academic offerings is over, students complete a survey comparing their levels of initiative, interaction, course outcomes, satisfaction, and information retention to see how the two groups fared. Using a step-by-step score evaluation approach, the questionnaire options are all quantified using scores with a range of 1–10 points or 1–5 points. At the same time, students in both groups were asked to fill out subjective questionnaires outlining their ideas for how the teaching techniques could be improved. After the completion of the regular academic offerings and the completion of the first survey questionnaire, all students in group A received augmented reality teaching (Group C). After completing all learning content, students in group C received the second survey. Two orthopedic experts reviewed the data and findings from the survey and used statistical methods to conclude. The flowchart of this study is shown in Figure 4.

Original data were analyzed using MedCalc 15.2.2 software (MedCalc, Ostend, Belgium). To check if the data follows a normal distribution, the Shapiro–Wilk test was applied. However, considering the particularity of questionnaire surveys, this study used non-parametric tests. The Mann–Whitney independent sample test was used to verify the correlation between the traditional teaching group (Group A) and augmented reality-assisted teaching group (Group B), and the data are described as the median and 25 and 75th interquartile ranges (Q25, Q75). U value, p value, and r values have been reported. The data are described as the median and 25 and 75th interquartile ranges (Q25, Q75); The effect size (ES) was calculated by dividing the Z-score by the square root of n: 0.20–0.49 is considered a small effect; 0.50–0.79 is considered a medium effect; and 0.80 and above are considered a large effect (15, 16). The Wilcoxon rank-sum test was used to verify the correlation between the traditional teaching group (Group A) and continuing AR teaching in the traditional group (Group C), and the data are described as the median and 25 and 75th interquartile ranges (Q25, Q75). Z-value and p values have been reported. p < 0.05 was identified as statistically significant. The Wilcoxon rank-sum test were created using SPSS 23.0 (IBM, Armonk, NY, United States). Figures were created using GraphPad Prism 10.0 (GraphPad Software, San Diego, CA, United States).

The traditional teaching group included 13 students, four males and nine females, who were all eighth-year clinical medical students in fifth grade or sixth grade. The average age of the participants in the traditional teaching group is 23.5 ± 0.9 years old. Twelve medical students (seven males and five females) from fifth grade or sixth grade in the augmented reality technology teaching group served as comparisons. The average age of the students in the augmented reality assisted teaching group is 23.7 ± 1.3 years old. In terms of age, stage of cultivation, etc., no statistically significant difference was found between the two groups. According to the research requirements, all participants fulfilled the inclusion criteria and finished the course, questionnaires, and post-class feedback assessments.

During this study, all students were taught theoretical concepts using slides and models of the human body. The augmented reality group then had percutaneous vertebroplasty surgical operation simulations and instruction using augmented reality models after receiving traditional instruction (Figures 5, 6). Teaching quality, students’ learning interest, anatomical structure mastery, and knowledge of percutaneous vertebroplasty were all significantly different between group A and group B, according to the results of the post-class evaluations. There were notable benefits (p < 0.05) for the group that was taught using augmented reality technology (Table 1). In addition, after finishing the regular academic offerings and the first teaching scenario survey, every participant in the traditional teaching group was also allowed to learn by AR simulation technology. A second survey with questions was administered following the completion of all training in the traditional teaching group (Group C). After implementing AR into the traditional teaching group, students demonstrated greater engagement in learning, enhanced comprehension of anatomical structures, and higher ratings of the overall quality of instruction (Table 2). Along with their proficiency in percutaneous vertebroplasty, students demonstrated considerable enhancements in their comprehension of the patterns of growth and spread of spinal malignancies (Table 2). The box plots were used to display the distribution among three groups (Figure 7).

According to the study above, the researchers collected the identified issues, methods for enhancement, and the teaching pros and cons for each set of test students by utilizing subjective feedback during the teaching process (Table 3). The survey questionnaires for each group are listed as attachment forms (Supplementary materials 1, 2). The findings indicated that using AR holographic glasses might lead to a higher incidence of discomfort among students, such as dizziness, while initially using them, primarily due to the weight of the helmet and the three-dimensional image. However, after making the required adjustments, discomfort can be quickly relieved.

Virtual reality (VR) technology has been instrumental in enhancing psychological skill training in high-performance professions, especially those that require a high level of safety awareness and complex skills (17). VR can efficiently train and enhance psychological skills by creating a safe, repeatable, and fully controllable training environment. In high-pressure and high-performance environments, the VR technology offers various innovative solutions for the training and learning of complex skills, which are often expensive or dangerous to practice in reality (18, 19). Extended reality (XR) is an advanced version and expansion of VR technology, and the utilization of XR in orthopedics and medical education is gaining popularity. AR technology emerges from integrating human visual exploration and interaction with computer-generated 3D worlds, which are then overlaid onto the real-world environment, incorporating virtual circumstances (20). In medical education, AR tools allow for anatomical simulation, evaluation, diagnosis, and surgical teaching and training through realistic surgical simulations (21). In recent years, the teaching attempts and applications of AR technology in undergraduate, graduate, and junior resident physicians have received immense attention (22).

Through the VR/AR technology, realistic operating environments can be simulated, enabling trainees to improve key skills in virtual contexts, such as attention, decision-making ability, situational awareness, adaptability, and stress management (18). Such signals can be monitored and evaluated by analyzing physiological parameters such as heart rate, respiratory rate, and electroencephalogram (EEG) (19, 23). By wearing AR glasses, students and teachers can learn about the anatomy, imaging knowledge, and surgical approach selection of relevant cases in space of virtual and real fusion. Applying this teaching method allows for a more intuitive and clear display of the anatomical structure of limbs, joints, and spine. Küçük et al. (24) found that the use of AR technology to learn anatomical knowledge has an impact on the academic performance and cognitive aspects of medical students. Lopes et al. (25) divided 20 medical students without any experience in basic surgical skills into an AR telestration teaching group and a traditional teaching group, and then evaluated the students’ learning experience, self-confidence, and self-assessment through questionnaires. Through result analysis, it was found that using helmet devices with AR functions can replace traditional teaching methods in terms of the quality of the surgical gesture, tension of the suture, self-evaluation or confidence. In addition, Nagayo et al. (26) established a self-training system that provides 3D information through AR technology. There was no significant difference in the improvement of test scores between the AR group and the video group, however, the actions provided by the AR system were more helpful in manipulating surgical instruments than videos (p = 0.02). Students can quickly grasp and apply knowledge points such as complex fracture, joint disorders, and spinal diseases. Meanwhile, it also frees medical education from the limitations of traditional teaching venues (22).

Surgery for tumor in bone and soft tissue is a specialized area that combines the expertise of orthopedics and oncology. Subspecialty teaching encounters numerous hurdles, including intricate anatomical underpinnings, a wide range of disease classifications, and lengthy training periods for young physicians. Although the application of VR/AR/MR technology in orthopedic specialties such as spine, joints, and trauma has been on the rise (27, 28), few studies have explored the application of VR/AR/MR technology in bone and soft tissue tumors (Table 4). The management of bone tumors requires long-term and continuous clinical learning and practical operation. Therefore, the VR/AR/MR technology may provide simulated clinical teaching incorporating hearing, vision, and touch elements, enabling students to participate in clinical practice under various scenarios. It also facilitates real-time guidance from teachers, increases teacher-student interaction, and improves teaching efficiency.

Spinal tumor surgery poses significant challenges and risks, particularly in cases of malignant tumors, when determining the tumor boundaries is typically challenging. The surgical procedure is challenging and demands exceptional surgical expertise from the operator. In this context, spinal tumors, as one of the most difficult bone tumor diseases in orthopedics, have the characteristics of difficult-to-grasp anatomical structures, difficult-to-understand tumor invasion characteristics, and long surgical learning curves. As an innovative teaching method, AR technology has rapidly developed with unique advantages such as immersion, interactivity, and accuracy, helping to promote the transformation and progress of teaching models in bone and soft tissue tumor surgery. To assure surgical safety, AR technology can visualize crucial anatomical structures, including main blood arteries, nerves, spinal cord, and other vital tissues. The time needed for surgical procedures can be significantly decreased by projecting these rebuilt anatomical structures onto the patient (47). Under such circumstances, this study explores the training effectiveness of applying augmented reality technology in simulated teaching of spinal tumors.

Cho et al. was the first to report the application of AR-assisted navigation technology in surgical resection of pig femoral and pelvic tumors, laying the foundation for the application of AR technology in the field of bone and soft tissue tumors (46, 47). Subsequently, Carl et al. (45) and Tigchelaar et al. (36) applied AR technology to the surgical treatment of spinal intradural tumor and spinal epidural tumor, respectively. However, although there are many literature reports, the expansion and attempts of AR-assisted teaching methods for medical education are especially lacking. In this context, our team developed an AR-assisted spinal tumor-teaching model and surgical teaching system for medical students.

This study was the first reported attempt to apply AR technology in medical teaching, using spinal tumors and percutaneous vertebroplasty as examples, head-mounted augmented reality devices were used by students to create specific learning scenarios, presenting the complex three-dimensional spatial structure intuitively. After further data analysis, we found that the AR-assisted teaching group had better teaching quality in learning interest, understanding of spinal analysis, understanding of spinal trabecular structure, understanding of the common sites for spinal tubers, and understanding of the general steps of PVP surgery, compared to the traditional teaching group. In addition, after receiving AR-assisted teaching in traditional teaching group, significant improvements had been found in teaching quality, learning interest, understanding of basic spinal anatomy, understanding of spinal trabecular structure, understanding of the common sites for spinal tumors, understanding of the involvement and metathesis of spinal tumors, understanding of the general steps of PVP surgery, and expectations for further operational simulation training. The above research results confirmed the teaching significance of the spinal tumor model constructed in this study and the demonstration model of percutaneous vertebroplasty surgery, as well as the important value of augmented reality technology in assisting spinal tumor teaching. The research results of this study will lay a theoretical and practical foundation for the application of AR-assisted spinal tumor anatomy teaching and preclinical teaching, as well as the improvement of medical education level.

Spinal tumors are a difficult point for medical students or young physicians during learning, most of them require prolonged learning period to master. The use of AR technology to simulate spinal tumor models and interactive surgical scenarios needs to be further verified in the medical preparatory stage or young physicians. For this purpose, our research team independently designed a spinal tumor model and a percutaneous vertebroplasty surgery procedure that can be used to achieve interactive teaching. Although the number of medical students included in the two groups was not high, statistically significant differences were observed between multiple indicators among groups. Future research will focus on expanding the participant pool, exploring diverse teaching methods and scenarios, and incorporating objective skill assessments (e.g., surgical operation simulations) with scoring systems. As the number of students increases, more teaching data will be accumulated for more in-depth analysis of multiple factors such as gender, age, student type, and major. In terms of device improvement, students will have more opportunities for interactive operations to enhance their clinical operational skills. These advancements will further validate the feasibility of AR as a valuable tool for modern orthopedic and surgical oncology education.

Machine learning and artificial intelligence technology can be applied in psychological skill assessment and training, and the combination of the two is expected to further improve the quality of medical education (18, 23). For example, using machine learning algorithms, patterns can be extracted from a large amount of physiological and behavioral data to evaluate individual performance and progress at different training stages (23). The technologies can also be used to establish adaptive learning systems that dynamically adjust training content and difficulty based on real-time feedback from trainees, thereby improving training effectiveness. In addition, artificial intelligence can create virtual coaches and intelligent decision support systems, which may improve the experience of trainees in virtual reality learning environments (49, 50). Thus, the promotion and improvement of AR technology in medical teaching need to accumulate more practical data and still face many challenges (51, 52).