The usefulness and potential of high-fidelity three-dimensional models in spine surgery training: cross-sectional empirical study

Abstract

Background

Advancements in understanding spinal disorders and diagnostic techniques have increased the range and complexity of spinal surgeries. However, constraints have arisen in gaining experience techniques through actual surgical cases due to considerations of medical safety, efficiency in working hours, and cost-effectiveness. As such, off-the-job training is expected to play an increasingly significant role. Three-dimensional models have been used for organizing knowledge and training surgical techniques. Their strengths lie in safety, accessibility, and cost-effectiveness. However, their proximity and limited realism make them less suitable for advanced training, restricting their use mainly to beginners. This study evaluates the potential of more realistic models for comprehensive surgical training and explores further applications of them.

Methods

MRT², a detailed three-dimensional model developed from real patient computed tomography (CT) data, offers realistic external characteristics and compatibility with radiological imaging. Three types of spine implant surgery seminars were conducted using MRT²: (A) cervical pedicle screw placement with fluoroscopy and navigation, (B) a comprehensive mock patient case—from understanding pathology to surgical planning and execution, and (C) lectures and hands-on deformity correction practice for various spinal conditions. Participants evaluated the models and seminars through questionnaires.

Results

Seminar A, comparing MRT² to a conventional model (Sawbones), found MRT² significantly more realistic in visual and performance realism, as well as tactile feedback (visual realism: MRT² 45, Sawbone 26, p = 0.0009; performance realism: MRT² 42, Sawbone 17, p = 0.0001; tactile feedback: MRT² 40, Sawbone 18, p = 0.009). In Seminar B, MRT² provided an immersive experience even for spine surgery specialists, closely mimicking clinical practice (Questions 14–18 regarding psychological aspect, scoring 18–19 out of 20). Open-ended responses noted MRT²’s unique benefits, such as allowing multiple participants to perform the same procedure for comparative planning and outcomes. Observing vertebral movements during corrective maneuvers further confirmed its educational value.

Conclusions

Enhanced structural detail and realistic simulation make these three-dimensional spinal models highly effective for both novice and specialist training, significantly improving the training experience across skill levels.

Background

With advances in understanding the pathology of spinal disorders and the progress in imaging diagnostics, the number of conditions eligible for spinal surgery has increased. In response to the demand for better treatment outcomes, surgical procedures have become more complex and sophisticated, supported by rapidly evolving advanced technologies [1]. Consequently, the skills and knowledge that spinal surgeons need to acquire are continuously expanding [2]. However, due to factors such as medical safety, revisions to work style and hours, and the need for operational efficiency related to profitability, there are growing limitations on young spinal surgeons’ opportunities to gain hands-on experience and mastery in actual surgical cases.

Furthermore, the acquisition of advanced and specialized skills is strongly associated with the time dedicated to deliberate practice—a form of practice that focuses on specific aspects of performance, involves concentrated effort, and incorporates immediate feedback, as described by Ericsson [3]. However, traditional surgical training, which revolves around practice and apprenticeship, often offers limited opportunities for such focused learning [4]. In this context, off-the-job training is expected to become increasingly important hereafter. Various methods are available for surgical skill training using simulations, including 3-dimensional(3D) models, animals, cadavers, simulators, and virtual reality [4, 5]. 3D Models are inexpensive, easy to obtain, can be replicated for repeated use, and offer tactile feedback that virtual reality cannot provide. They also offer safety advantages compared to animal models and cadavers. On the other hand, models often lack the realism due to their appearance and familiarity and can result in lower immersion levels during training. This shortcoming may lead trainees to handle the models less carefully, making their techniques or overall training experience less precise and somewhat careless. For this reason, model-based training is often considered suitable only for beginners or basic procedural training [4].

Therefore, in this study, we hypothesized that, if precision and realism is pursued, spinal models could serve as a valuable training resource not only for beginners but also for young spinal specialists, and we aimed to verify this hypothesis.

Methods

A simulated surgery was performed using a high-fidelity three-dimensional spinal model compatible with X-ray examination. This model, called MRT², is a polyurethane foam model created from patient Digital Imaging and Communications in Medicine (DICOM) data. It mimics the two-layer structure of cancellous and cortical bone by combining two selected types of urethane foam with different densities and hardnesses, which were optimized after repetitive biomechanical tests. Additionally, its X-ray-opaque surface treatment allows for imaging with X-ray radiography and X-ray CT scans. (Fig. 1) Using this 3D model, we conducted three surgical training sessions for orthopedic residents and spinal surgeons. Based on participants’ feedback, we evaluated the potential and usefulness of 3D models in spinal surgery training.

Fig. 1

MRT². (a) Exterior of MRT2. (b) Axial slice of the model of the C7 vertebra, showing its two-layer structure representing the cortical and cancellous bone. A casting mold for both cancellous and cortical bone is created from a 3D printer master based on a patient’s CT scan DICOM data. The cortical bone portion is made from these casting molds for each vertebral body, and urethane for cancellous bone is filled inside. As a finishing touch, a zinc coating is applied to the surface of the vertebral body. The hardness of the urethane used for both cortical and cancellous bone was determined through repeated mechanical testing to identify the optimal grade. (c, d) X-ray images of MRT2. The radiopaque surface allows it to appear clearly on X-ray films and CT scan images. (e) The appearance of MRT2 on the CT Navigation monitor

Three types of training seminars of simulated surgeries were performed:

• A) Cervical Pedicle Screw Placement Training seminar.

A workshop for the procedure for inserting cervical pedicle screws using lateral fluoroscopy and CT navigation was held in our operation theater. The ten participants were the members of our university orthopaedic department or the medical students rotating the department. The procedure was carried out on both the MRT² model and a traditional skeletal model (Sawbones, A Pacific Research Company, USA). (Fig. 2)

Fig. 2

The procedure was carried out on both the MRT² model and the Sawbone model. (a, b, c, d) MRT², and (e, f, g, h) Sawbone. (a, e) Setting of the model with navigation equipment. (b, f) Image of X-ray intensifier, (c, g) View on the navigation monitor, (d, h) CT scan image after screw placement. The MRT2 model is more clearly visible on X-ray and CT scan images. On the navigation monitor, the two-layer structure of the MRT2 appears more realistic compared to the Sawbone model

A questionnaire from a previously published survey by Leblanc was translated and modified for spinal surgery and was used to evaluate the surgical training using these models, as well as the ease of use of the models themselves [6]. Participants rated each question on a scale of 1 to 5 for both models, and the results were statistically compared using the Wilcoxon rank-sum test. Statistical analysis was performed via JMP Pro 18.0.1 (JMP Statistical Discovery LLC. USA).

• B) Simulation of Diagnosis and Treatment on a Mock Patient.

A mock patient was created using real patient data, and participants were required to diagnose and treat (perform surgery) in a structured program. Four young spinal surgeons at PGY 7 to 10 of the spine division of our orthopaedic surgery department participated. A mock patient was created in the hospital’s electronic medical record system, with chief complaints, present illness, physical findings, and various imaging tests prepared. (Fig. 3)

Fig. 3

Resources for session B: simulation of diagnosis and treatment on a mock patient. (a) A mock patient was created in the hospital’s medical chart system. Participants accessed the chart using the patient ID to retrieve and review clinical information. (b) Necessary imaging studies, including X-ray films, CT scans, MRI, and myelography, were made available, replicating daily clinical settings. (c, d) Extension and flexion X-ray films of the bone model were also accessible through the PACS system. The bone model, embedded in a polyurethane mold, was designed with some degree of flexibility. (e) The simulation surgery was performed in the operation theater, and the equipment and surgical instruments necessary for the surgery was prepared including CT navigation and fluoroscopy. (f) During the surgical procedures, the model was covered with a surgical drape

The images were anonymized clinical test images from real patients. Participants collected clinical information from the mock patient’s electronic records, diagnosed the pathology and etiology independently, and formulated a treatment plan and detailed surgical technique to be submitted within a month. Based on the submitted surgical plans, necessary instruments, fixation devices, and imaging support equipment were prepared in the operating room. A highly detailed 3D model (MRT²) created from the mock patient’s imaging data was embedded in polyurethane to simulate soft tissue, and an operative field was created over it. Six weeks after obtaining the mock patient information, participants performed the surgery according to their devised plan in the operating room. Postoperatively, plain X-ray and CT scans were performed to verify the placement of implants and correction outcomes. Feedback on this simulation and the model used was collected via a questionnaire. The same questionnaire as in Experiment-A was used.

• C) Paid Hands-on public surgical skill training seminar held by the Spine Surgery Society.

A Japanese spine surgery society organized a hands-on course on surgical techniques using 3D spinal models, targeting young spine surgeons. Participants were recruited at a participation fee of 20,000 yen. After several lectures on the pathology and surgical techniques for cervical spondylosis, idiopathic scoliosis, and lumbar degenerative scoliosis, participants practiced these procedures on disease-specific models. Feedback on the course and the models used was collected via a unique questionnaire different from the previous two.

All questionnaires used a 5-point Likert scale from 1 to 5 (1: Strongly Disagree − 5: Strongly Agree) along with free comments. Participants completed the questionnaire immediately after the practice sessions, and responses were collected on the same day.

Results

Cervical pedicle screw placement training seminar

Ten participants were included in this seminar. Five participants were spine surgeons with post-graduate years (PGY) 8 or higher who specialized in spinal surgery after completing general orthopedic training, three participants were orthopedic residents at PGY 3 to 6, and two were medical students. Responses were received from all ten participants for all the questions. (Table 1)

Table 1 Questions and the sum of the points from ten participants. (p value < 0.05 is boldface.)

For questions about visual realism, performance realism, and tactile feedback, and realistic sensation of the procedure (Questions 3,5, 8, 11, 15), MRT² acquired statistically significantly better score. In questions regarding the effectiveness of models for learning surgical techniques such as Questions 20, 21, 22, MRT² received significantly higher scores, suggesting that the precision of the model is important in this regard. The experience of screw placement using the models received high ratings for both models　(Question 31).

Their answers were compared and assessed between two groups: senior group of five senior spine surgeons and junior group of five orthopedic residents and interns. Those questions with significant difference in the sum of the five participants’ scores between the two groups are shown in Table 2.

Table 2 The questions which had statistically significant difference between the experienced seniors and the junior participants, and the sum of the scores from 5 seniors and 5 juniors are shown. No question had significant difference in both sawbone and MRT² at the same time

In general, the seniors gave lower scores than the juniors, especially for the Sawbone model. The seniors gave a particularly harsh evaluation of Sawbone model in terms of realism. (questions 15, 19). In the MRT² model, junior surgeons scored full points on several questions, showing a significant difference from senior surgeons’ ratings. These questions were for the effectiveness of the model as the device to provide surgical experience (Questions 22, 24, 26).

Simulation of diagnosis and treatment on a mock patient

The case prepared was a global malalignment lady with cervical and lumbar kyphosis. All four young spine surgeons, who routinely performed spinal decompression and short-segment decompression and fusion surgeries but had never served as the primary surgeon for spinal deformity correction surgery, diagnosed her primary pathology as cervical kyphosis and planned two- or three-stage anterior-posterior corrective fusion surgeries. (Table 3)

Table 3 The model used in the simulation on a mock patient. The surgical plan, time spent for the mock surgery, and the postoperative X-ray by four participants are shown. All the participants spent long hours for the pre-procedural Preparation and treated the neck model very carefully

Three participants used lateral fluoroscopy during surgery, while one used CT navigation. On average, participants spent 4.1 ± 2.1 h (range: 2–7 h) on preoperative learning, 1.3 ± 0.65 h (range: 0.5–2 h) on surgical preparation/planning, and 127 ± 17 min (range: 105–145 min) on the surgery. Survey responses showed that all participants rated the model’s appearance (Question 4) and the accuracy of the drilling (Question 10) highly, with all giving a score of 4 or higher. (Table 4)

Table 4 The answers of the four participants (A, B, C, D) in the simulation on a mock patient. Answers 1 = strongly disagree, 2 = disagree, 3 = neutral, 4 = agree, 5 = strongly agree

Likewise, for questions on the immersiveness and realism of the procedure, all participants scored 4 or higher, suggesting the model’s utility in the surgical training even for those who perform spine surgery in their daily practice. Three out of four participants strongly believed that this type of simulation should be integrated into clinical training programs.

Paid hands-on public surgical skill training seminar held by the spine surgery society

Eight participants attended the course. Responses were obtained from seven out of these eight participants and two of the instructors, totaling nine responses. The breakdown was as follows: eight orthopedic specialists and one non-specialist, three spinal surgery instructors from the Japan Spine and Spinal Cord Society (JSSR), and six non-instructors. In terms of experience in spinal surgery, two participants had over 15 years of experience, one had between six and ten years, and six had less than five years of experience.

Answers to the questions are presented in Table 5.

Table 5 The questions for the hands-on surgical skill training seminar. The number of the people per each point are shown. The seminar and the model used in the course were generally accepted very positively

The model’s appearance and the tactile feedback during probing and screw placement were felt realistic by seven participants. All eight participants agreed that the model was an effective method for learning spinal implant surgery, and that combining practical training with a model enhances understanding of surgical strategy and techniques compared to theoretical learning alone. Regarding the seminar, six participants found it very useful, while three found it useful. When asked about the likelihood of recommending a similar surgical training course to colleagues or acquaintances, seven out of the nine participants rated it 9 or 10 on a scale of 0 to 10 (0 = not at all, 10 = extremely likely), yielding a Net Promoter Score (NPS) of 78.

Discussion

In surgical training, various hands-on techniques are used, such as models, animals, cadavers, and virtual reality [4, 5]. Among these, training with models is particularly noted for being affordable, safe, and convenient. Studies have reported that combining model-based practice can improve and solidify learned techniques [7, 8]. Furthermore, in a comparison of understanding complex structures between 3D images and 3D models, it has been shown that using 3D models enhances comprehension [9, 10]. Meanwhile, 3D model training tends to be handled with less care due to its lack of precision and immersion and overly familiar nature and it has been considered useful only for simple purposes such as confirming basic techniques or providing explanations for beginners, and unsuitable for advanced training [4]. In this study, we aimed to explore whether enhancing the precision and realism of models could expand their potential as a hands-on training tool. The models used in this study mimicked both cortical and cancellous bone with two layers, and the surface was coated with a radiopaque material, allowing for clear evaluation in X-ray imaging. These models were created by 3D printing based on real patient image data.

In the Seminar A, participants, ranging from beginners to the professor of spine surgery, compared traditional models with precision models during cervical pedicle screw placement training using X-ray imaging and CT navigation. The precision model was rated as being closer to reality in terms of appearance, tactile feedback during the procedure, and the visibility in X-ray imaging. This confirms that the precision model (MRT²) demonstrated higher performance than traditional models as a tool for procedural practice. Although there was no statistically significant difference between the models in terms of the perceived value of screw placement training, both models received positive feedback, indicating that model-based procedural training itself is beneficial regardless of precision. In a sub-analysis based on surgical experience, senior doctors generally assigned lower scores to both models. While senior doctors did not rate the detailed models poorly, junior doctors provided significantly higher scores, particularly in terms of the models’ utility for practicing surgical techniques. From this perspective, although surgeons with a broad range of experience found the detailed models more useful, it appears that as surgical experience increases, there remains a limit to the training depth achievable through model-based simulations.

In the Seminar B, we provided a more realistic environment to allow young spine surgeons to experience a simulated surgery on a challenging case. The surrounding environment was designed to resemble a real operating room, complete with patient information from electronic medical records and standard surgical instruments. For participants who had never performed surgery for spinal deformities, the subject matter—a challenging case of cervical kyphosis—was particularly difficult. However, each participant invested significant time in pre-operative study and surgical planning, with surgeries lasting an average of over two hours, during which they worked with meticulous attention to detail. One participant noted, “I realized valuable things by stumbling over aspects I hadn’t considered pre-operatively.” This comment highlights how seriously the participants approached the simulation, allowing them to recognize finer points as if they were dealing with an actual surgery. One common criticism of model-based training is that it can lead to careless or sloppy technique. However, this was mitigated by designing a realistic environment, allowing participants to stay focused and engaged. Feedback from the participants, all of whom were early-career spine surgeons, highlighted the effectiveness of realistic model-based surgery in learning and practicing surgical techniques, as well as in experiencing complex surgeries—especially for procedures they had only studied in literature and textbooks, but had never or rarely performed in a real clinical setting.

In the Seminar C, we opened the course to the public and held a paid surgical training session. The course covered both classroom lectures and hands-on practice using disease-specific models for conditions like adult spinal deformity and idiopathic scoliosis of the cervical and lumbar spine. Participants rated the performance of the models highly. The combination of lectures with hands-on training using precision models was seen as deepening understanding and facilitating immediate application to real cases. Regarding costs (a ¥20,000 participation fee and travel expenses) and the seminar’s duration (about 8 h), more than half of the participants felt they were appropriate, and no significant complaints were raised about the burden.

In this study, we examined the potential and scalability of precision models across various training subjects. By enhancing both the precision of the models and the surrounding environment, we were able to provide a realistic training experience that was not only suitable for beginners but also highly satisfactory and immersive for spine surgery specialists.

According to Cognitive Load Theory, learning effectiveness can be discussed by categorizing the cognitive load during learning [11, 12]. This theory suggests that by reducing unnecessary load (extraneous load) and keeping the overall cognitive load within the limits of the learner’s cognitive resources, learners are less likely to experience difficulties in task performance and generally demonstrate higher performance outcomes. Pollock’s research on assimilating complex information indicates that when learners are exposed to a large amount of information from the start, their understanding is lower compared to when information is initially limited to isolated elements that can be processed serially [13]. To regulate cognitive load and ensure it stays within the learner’s capacity, adjusting the amount of information presented based on the learner’s level is essential. Manipulating the fidelity of the learning environment is a way to gradually increase the number of interacting elements [12]. In this regard, we found that the surgical training using these detailed three-dimensional models, with adjustable fidelity, can serve as excellent training methods for a wide range of learners, from beginners to specialists.

Limitations of this study include the small number of participants in each practical exercise, which limited the number of responses we could analyze. In the Seminar A and B, participants came from the same department as the researchers and answered questionnaires with their names attached, raising concerns about potential bias. However, in the Seminar C, which was open to the public and used anonymous surveys, the precision models received similar high evaluations as in the Seminar A and B. Additionally, since the Seminar B and C used only a single model without comparative analysis, we could not evaluate the impact of model precision or compare it to other training materials. Furthermore, because we focused on participants’ subjective feedback, we were unable to assess the models’ effectiveness in terms of knowledge retention or procedural mastery.

Conclusions

Spinal 3D models, when enhanced for precision and paired with a well-designed environment, can serve as highly satisfying and effective training tools not only for beginners but also for spine surgery specialists.