Educational Utility of a Full-Skull 3-Dimensional Model for Skull Base Surgery Training Authors: Daisuke Kuga, Koji Yoshimoto Published: 2026 ABSTRACT BACKGROUND AND OBJECTIVES:Skull base surgery requires precise 3-dimensional (3D) anatomic understanding and advanced drilling techniques within a confined operative space. Although cadaveric dissection remains the gold standard for training, its availability is limited. This study aimed to evaluate the educational utility of a commercially available full-skull 3D model for skull base surgery training.METHODS:A full-skull 3D model reproducing the cranial base and brain was used in a hands-on training session involving 8 neurosurgeons, including 2 residents and 6 board-certified surgeons. Multiple skull base approaches were practiced. In addition, a patient-specific tumor model generated from clinical imaging data was incorporated to simulate pathological anatomy. Participants completed a questionnaire using a 5-point Likert scale to assess bone hardness, anatomic accuracy, realism, drilling sensation, safety and usability, and overall satisfaction.RESULTS:Overall satisfaction with the model was high across all evaluation categories. Anatomic accuracy, realism, and usability received high mean scores. Overlay visualization of individual responses demonstrated consistent trends without reliance on inferential statistics. Qualitative feedback indicated that the model was particularly useful for understanding spatial orientation and approach selection, whereas limitations were noted regarding tactile realism, including uniform bone hardness and dural texture.CONCLUSION:The full-skull 3D model demonstrated substantial educational value as a supplementary training tool for skull base surgery. By enabling repeated practice, approach comparison, and patient-specific simulation, this model may complement cadaver-based training, particularly for early-stage neurosurgeons. Further studies with larger cohorts are warranted to establish its role within structured skull base education programs. Skull base surgery demands precise operative maneuvers within a confined space where critical neurovascular structures and sensory organs are densely clustered in deep regions. Accordingly, an accurate 3-dimensional (3D) understanding of the regional anatomy and the acquisition of bone-drilling techniques based on this anatomic knowledge are essential.1 However, 2-dimensional imaging modalities such as computed tomography (CT) or MRI alone are insufficient for fully appreciating the complex spatial relationships and stereoscopic configuration of the skull base.2 Therefore, hands-on surgical training using models that closely simulate actual operative procedures is indispensable.2 Cadaveric dissection represents the educational modality most closely resembling real surgery, providing authentic tactile feedback of bone and soft tissues and detailed neurovascular anatomy.3 It remains an indispensable tool for mastering skull base anatomy. Nevertheless, cadaver-based training is associated with substantial challenges, including ethical and legal requirements, maintenance of preservation facilities, disposal procedures, and considerable operational costs. Furthermore, religious or cultural circumstances may preclude the use of cadaveric materials in some regions, making it difficult to establish environments where young surgeons can practice repeatedly. In recent years, 3D-printed artificial skull models have gained attention as educational tools that may compensate for the limitations of cadaveric training. Artificial skull models are easy to obtain and prepare, do not require the ethical procedures associated with cadaver use, and offer advantages in safety and cleanliness during drilling.4,5 Because their use does not require specialized facilities such as air-conditioned laboratories, they can be used in various settings and are considered suitable for repetitive practice by junior trainees. However, several limitations have been noted, including differences in bone texture compared with real human bone and inferior reproduction of fine anatomic details relative to cadaveric specimens. The KEZLEX® produced by Japan Medical Company is one of the widely used artificial skull models in Japan. This model is characterized by high anatomic fidelity as it is generated from actual human CT and MRI data sets, enabling detailed replication of real skull base anatomy.6,7 Nonetheless, no detailed reports have evaluated its educational utility or limitations in comparison with cadaveric specimens or other commercially available models. In this study, we examine the educational value of the KEZLEX® based on its structural characteristics. Furthermore, through comparisons with cadaveric materials and other artificial models, we assess the usefulness of this model for surgical training in the field of neurosurgery. METHODS This study was approved by the Institutional Review Board of our institution, including the use of clinical data and radiological and intraoperative images for the creation and evaluation of the patient-specific tumor model. Written informed consent for the surgical procedure and for the use of the patient's clinical data and radiological and intraoperative images for research and publication was obtained. Skull Model Used in This Study The 3D full-skull model used in this study (KEZLEX®; Japan Medical Company) was developed as an extension of the dissectible skull base model reported by Mori et al and Yamaoka et al.8,9 While these models focused on partial skull base anatomy, the present model reproduces the entire cranial structure, encompassing the anterior, middle, and posterior cranial fossae. Major intracranial arteries, cranial nerves, and dural venous sinuses are incorporated, allowing multiple skull base approaches to be practiced sequentially within a single integrated platform. The cranial framework is fabricated from a rigid synthetic material selected to provide stable handling and reproducible resistance during bone drilling, suitable for repeated surgical training. In accordance with previously reported layered dural models, the dura mater is simulated using polyester fabric. Two differently colored layers are superimposed to represent the bilaminar structure of the native dura mater, enabling visual discrimination of dural layers and facilitating dural incision and dissection maneuvers.8 In contrast to skull-only models, this full-skull system additionally incorporates a deformable, sponge-based brain component. This feature allows simulation of gentle brain retraction and dynamic changes in the operative corridor, thereby providing a spatial environment that more closely approximates actual skull base surgical conditions (Figure 1A and 1B). Overview of the full-skull model and the patient-specific tuberculum sellae meningioma model. The full-skull training model is shown in A, an anterior view and B, a lateral view. Portions of the temporal bone are removed to visualize internal structures, and a sponge-based brain parenchyma allows simulation of brain retraction and operative manipulation. C-E, A patient-specific tuberculum sellae meningioma model: The tumor was constructed using green modeling clay (black arrow). C, The original radiological images used for model construction, D, a reconstructed surgical view, and E, an oblique superolateral view. The optic nerve and anterior cerebral artery are displaced superiorly by the tumor, reproducing surgical anatomy. Another characteristic of this model is its compatibility with patient-specific pathology. Tumor replicas generated from individual patient radiological data sets can be integrated into the cranial cavity, enabling the model to be used not only for anatomic education and approach training but also for case-specific preoperative surgical simulation. Creation of the Patient-Specific Meningioma Model and Intraoperative Comparison A patient with a tuberculum sellae meningioma who underwent surgery at our institution was selected, and the tumor replica was fabricated using CT and MRI Digital Imaging and Communications in Medicine data sets. The tumor morphology, spatial deviation of the optic apparatus and adjacent vascular structures, and dural attachment were reproduced as accurately as possible and fixed inside the cranial model (Figure 1C-1E). A simulated tumor resection was then performed using a supraorbital eyebrow approach. The operative field, visualization, and reproducibility of key anatomic structures were compared with corresponding intraoperative photographs from the same patient to evaluate realism and educational utility. Hands-on Course and Questionnaire-Based Evaluation A hands-on training session was conducted using the model with 8 neurosurgeons (2 residents and 6 board-certified neurosurgeons), all of whom had limited experience with complex skull base approaches. During the session, participants performed multiple skull base approaches using the model, including the anterior transpetrosal approach, combined transpetrosal approach, and transcondylar approach. Additional microsurgical procedures, such as anterior clinoidectomy after a frontotemporal craniotomy, were also performed. After the training session, participants completed a structured questionnaire evaluating 6 items on a 5-point Likert scale: (1) bone hardness, (2) overall realism of the model, (3) drilling sensation, (4) anatomic accuracy, (5) safety and ease of handling, and (6) overall satisfaction. Free-text comments were also collected (Table). TABLE.Questionnaire Items Used for Model EvaluationItem no.Evaluation domainQuestion1Bone hardnessThe hardness and resistance of the bone were appropriate for surgical training.2Model realismThe model realistically reproduced skull base anatomy.3Drilling sensationThe drilling sensation was similar to that experienced in actual surgery.4Anatomic accuracyAnatomic landmarks and spatial relationships were accurately reproduced.5Safety and usabilityThe model was safe and easy to handle during training.6Overall satisfactionOverall, I was satisfied with this model as a training tool.Data are presented as responses to a questionnaire using a 5-point Likert scale (1 = very poor, 2 = poor, 3 = fair, 4 = good, 5 = excellent). Questionnaire Items Used for Model Evaluation Data are presented as responses to a questionnaire using a 5-point Likert scale (1 = very poor, 2 = poor, 3 = fair, 4 = good, 5 = excellent). RESULTS Comparison Between the Patient-Specific Model and Intraoperative Findings Using the patient-specific tuberculum sellae meningioma model, a simulated tumor resection was performed through a supraorbital eyebrow approach. The simulated operative field and visualized structures demonstrated a high degree of concordance with the actual intraoperative observations. In particular, the spatial configuration of displaced structures—including the course and deviation of the optic nerve and its 3D relationship to adjacent vascular structures such as the internal carotid artery—was reproduced with favorable accuracy compared with intraoperative photographs. In addition, because the model incorporates a flexible brain component, it was possible to simulate dynamic changes in the operative corridor caused by brain retraction, as well as restrictions in working space—features characteristic of real skull base surgery. This represents an advantage over conventional skull-only models, which primarily provide static anatomic information without enabling realistic intraoperative maneuvering (Figure 2A-2E). Comparison of operative views between the patient-specific tumor model and actual surgery. The left supraorbital eyebrow approach is demonstrated. After gentle brain retraction, the region surrounding the left ON is inspected. A, The operative view obtained using an exoscope, and B, the corresponding view in the patient-specific tumor model. C and D, Post–tumor resection views obtained with the exoscope and the model, respectively. E, An endoscopic view acquired after insertion of the endoscope, allowing visualization of the posterior aspect of the optic nerve. The asterisk indicates the right internal carotid artery, and the black triangle denotes the right ophthalmic artery. ON, optic nerve. Questionnaire Results (n = 8) Eight neurosurgeons participated in the hands-on evaluation session, including 2 residents and 6 board-certified neurosurgeons. All participants had limited previous exposure to complex skull base approaches. Questionnaire responses were collected using a 5-point Likert scale. Overall satisfaction was high across all evaluation categories. In particular, model realism and safety and usability received high ratings, each with a mean score of 4.25, followed by anatomic accuracy with a mean score of 4.00, indicating a strong consensus regarding the educational value of the model in understanding skull base anatomy (Figure 3). Box-and-whisker plots of questionnaire results. Box-and-whisker plots show the distribution of scores for each questionnaire item. The central line represents the median, boxes indicate the interquartile range, and individual data points represent responses from each participant. Scores were recorded using a 5-point Likert scale (1 = very poor, 5 = excellent). Analysis of free-text comments revealed several recurring areas for improvement. Some participants reported that bone hardness and drilling sensation were relatively uniform and did not fully replicate the tactile feedback encountered during actual bone drilling. In addition, several respondents noted that the simulated dura mater felt firmer than expected and that its incision characteristics differed from those experienced in live surgery or cadaveric dissection. Despite these limitations, qualitative feedback was predominantly positive. Representative comments included statements such as “helpful for understanding approach orientation” and “facilitates visualization of anatomical relationships,” underscoring the model's usefulness for spatial and anatomic comprehension. Taken together, these findings indicate a high level of overall acceptance of the model as a training tool, while also identifying specific aspects—particularly tactile fidelity and material heterogeneity—that warrant further refinement. DISCUSSION Key Results In this study, the anatomic accuracy, operability, and clinical applicability of combining a full-skull 3D model with a patient-specific tumor model were confirmed, suggesting that this system represents a valuable training tool in skull base surgery education. A major advantage of this model is its ability to reproduce the entire skull, including the brain, rather than isolated cranial segments. The skull base forms a complex 3D continuum spanning the anterior, middle, and posterior cranial fossae, requiring surgeons to understand not only local anatomy but also global spatial relationships. This perspective is particularly important in cases such as large petroclival meningiomas that extend across multiple compartments, where a compartment-based anatomic understanding alone is insufficient.10 While conventional partial models are useful for region-specific training, they provide limited support for appreciating the skull base as a continuous anatomic structure. By contrast, the present full-skull model integrates the temporal bone, anterior skull base, middle cranial fossa, and posterior cranial fossa into a single platform, allowing sequential comparison of multiple surgical approaches—an important educational advantage. Furthermore, the inclusion of a brain component is a key feature of this model. Skull base procedures commonly require navigating the narrow space between the skull base and the brain, meaning that depth perception, access angle, and operative corridor awareness are essential elements of procedural mastery.11 These aspects are difficult to reproduce using conventional skull-only models. In comparison, the present model allows learners to access the lesion through the limited working space created by gentle brain retraction, providing a training environment that closely mimics real skull base surgical conditions. As a result, this model is considered highly useful for understanding the conceptual and technical differences among various skull base surgical approaches. The results of the questionnaire further support the educational usefulness of this model. Anatomic accuracy (mean score: 4.00), model realism (mean score: 4.25), and safety and usability (mean score: 4.25) received high evaluations, with many respondents indicating that the model was helpful in understanding the spatial relationships of critical structures and in selecting appropriate surgical approaches. Conversely, several participants noted limitations related to tactile realism, including uniform bone density, drilling feedback, and the feel of dural incision. These issues represent common challenges shared among artificial simulation models that do not incorporate biological materials and may be improved in the future through advancements in 3D-printing materials and multilayered manufacturing technologies.12-16 Interpretation and Educational Implications From an educational perspective, the use of a 3D skull model also provides meaningful advantages. Although cadaver dissection remains a highly valuable modality for learning skull base surgical techniques, cadaver-based training has inherent limitations, including specimen degradation, the need for specialized preparation and storage, and institutional or regulatory restrictions. These factors make repeated practice challenging, particularly for trainees who must acquire basic skills through iterative exposure.5,6,17 By contrast, the full-skull model used in this study is easy to prepare and allows learners to repeat identical procedures multiple times, enabling stepwise acquisition of essential surgical skills. This feature is particularly beneficial for novice neurosurgeons who are just beginning to train in skull base surgery as it lowers the threshold for practice and provides opportunities to rehearse different surgical approaches without risk. Next, the educational significance of incorporating a patient-specific tumor model must be addressed. The anatomic characteristics of skull base tumors—such as tumor location, bone sclerosis or erosion, and displacement of adjacent neurovascular structures—vary substantially among individual cases, and these differences directly influence surgical difficulty. Because cadaveric specimens represent normal anatomy, they do not replicate pathological anatomy and therefore cannot accurately reproduce the operative field or working corridor encountered during tumor resection. By contrast, the present model can be generated directly from the patient's radiological data set, allowing faithful replication of the tumor itself and the altered relationships between the lesion and surrounding cranial nerves and vessels. In this study, the spatial relationships visualized using the model demonstrated a high degree of concordance with intraoperative findings. This suggests that the model may play a valuable role in supplementing traditional cadaver training by providing opportunities to simulate tumor-specific surgical decision-making. A further advantage of the model is the ability to compare multiple surgical approaches for the same lesion, which is not feasible in real clinical settings. In practice, a surgeon cannot attempt multiple approaches in 1 patient, and thus, such a comparative understanding has historically depended on accumulated clinical experience rather than structured training.18,19 Using this model, however, different approaches—each based on the same patient data set—can be evaluated sequentially, enabling learners to directly compare operative exposure, accessibility, and working corridor characteristics across approaches. This facilitates a clearer understanding of approach-specific anatomic limitations and relationships to critical neurovascular structures. Such learning experiences are difficult to achieve through cadaver dissection alone or through conventional 2D imaging-based instruction. For trainees and early skull base learners in particular, the ability to simulate anatomic variation and evaluate multiple operative strategies represents a significant educational benefit, supporting the development of a structured approach to surgical planning rather than relying solely on experiential learning. An interesting observation from the free-text comments was that respondents did not simply request a more realistic model; rather, they suggested that the ideal level of realism depends on the educational objective and learner experience. Specifically, beginners seemed to benefit more from a uniform and easily drillable material, whereas more experienced surgeons expressed a preference for higher-fidelity models that more closely approximate the variability and technical demands of actual surgery. These findings suggest that customizing material characteristics and fidelity according to the trainee level may enhance the efficiency of skill acquisition.15 In other words, providing lower-fidelity, safety-oriented models for novice learners and high-fidelity models capable of supporting strategic planning and technical refinement for advanced learners may contribute to establishing a more structured and effective simulation-based skull base education system. Generalizability In regions where cadaver use is restricted because of cultural, ethical, or religious constraints, 3D-printed models may serve as a crucial educational alternative. Because this model is commercially available and its supply chain is stable, it has the potential to reduce disparities in skull base training environments and contribute to establishing more equitable surgical education systems globally. At the same time, the generalizability of the present findings should be interpreted with caution. The educational value of this model may vary depending on trainee experience, institutional training systems, available surgical equipment, and the specific skull base approaches being practiced. Further validation in larger and more diverse training environments is therefore needed to determine how this model can be optimally incorporated into structured skull base education programs. Limitations This study has several limitations. First, the number of participants was relatively small (n = 8), which may limit the generalizability of the findings. Second, the educational evaluation was based primarily on subjective questionnaire responses, and objective performance indicators—such as procedural time, accuracy of drilling boundaries, or error frequency—were not formally assessed. Third, the patient-specific simulation component was examined in a single representative case, and its applicability across a broader range of skull base lesions remains to be clarified. In addition, comparisons with cadaveric specimens and other commercially available models were based on conceptual and educational considerations rather than direct head-to-head experimental validation. Moreover, although the sponge-based brain model allows brain retraction using a retractor, it cannot reproduce intraoperative tissue injury, edema, bleeding, or other tissue responses caused by surgical manipulation. Furthermore, the stability and durability of this model when fixed with an actual head fixation device, such as a Mayfield frame, were not evaluated in this study. Future studies involving larger multicenter cohorts, objective assessment metrics, direct comparative analyses, evaluation of surgical positioning, and comparison with cadaveric training or other simulation models will be important to better define the role of this model in structured neurosurgical training. CONCLUSION The full-skull 3D model represents a versatile educational platform that enables anatomic understanding of the skull base, comparison of multiple surgical approaches, tumor simulation, and repeated hands-on practice within a single system. By addressing several limitations inherent to cadaver-based training, this model has the potential to support standardization of skull base surgical education and contribute to improved skill acquisition among early-career neurosurgeons.