Customized 3D-printed models for training in repairing large maxillary sinus membrane perforations Authors: Haoxin Lv, Yitong Sun, Yingqi Cai, Mingyu Zhao, Shici Xia, Yue Sa Published: 2026 ABSTRACT AbstractObjectiveThis study aimed to design and fabricate a personalized hard-soft tissue integrated 3D-printed model for hands-on training of large maxillary sinus membrane perforation repair and evaluate its training efficacy and cost-effectiveness to provide an efficient, standardized training solution for complex maxillary sinus-related surgical procedures in dental clinical practice.MethodsCustomized models replicating maxillary sinus anatomical structures were fabricated from clinical CBCT and intraoral scan data, using tissue-mimicking materials to simulate clinical scenarios. Twenty dentists with basic clinical experience but no prior hands-on experience in large maxillary sinus membrane repair completed standardized simulated surgery on the models. Training outcomes were assessed by pre- and post-training questionnaires, with data analyzed via the Wilcoxon signed-rank test, and the model’s cost-effectiveness was evaluated.ResultsParticipants’ pre- and post-training questionnaire scores differed significantly (p < 0.001). The 3D-printed model accurately replicated the clinical scenario of large maxillary sinus membrane perforation after cyst enucleation and helped trainees rapidly master key repair techniques. The total cost of training 20 participants was $315, with a marked economic advantage over traditional training models.ConclusionsThe personalized 3D-printed model accurately simulates the clinical scenario of large maxillary sinus membrane perforation after lateral window sinus floor elevation and cyst enucleation. It provides a safe, efficient, hands-on training platform for clinicians to master this complex repair technique, and with low cost, a customizable design, and high simulation fidelity, it serves as an innovative tool for standardized training of complex dental perforation repair, helping improve clinical repair and bone grafting success rates.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12909-026-09427-w.Keywords: 3D-printed models, Dental education, Lateral window sinus floor elevation, Maxillary sinus cyst, Large maxillary sinus membrane perforation repair, Simulation training Introduction Lateral window sinus floor elevation is a commonly used core technique for addressing bone deficiency in the maxillary posterior region [1]. During this procedure, maxillary sinus mucosal perforation is a frequent complication with an incidence rate of 10%–56%, among which large perforations (diameter > 10 mm) account for 19.23% of cases [2, 3]. Such defects are particularly common in the enucleation of maxillary sinus cysts. Notably, lateral window sinus floor elevation, maxillary sinus cyst enucleation, and postoperative large perforation repair form a coherent, highly technical procedural system [4]. For clinicians with limited clinical experience, performing this combined intervention directly without systematic hands-on training poses extremely high operational risks, which is why this integrated set of procedures demands proficient mastery in clinical practice. Traditional training tools struggle to cover this comprehensive clinical scenario: animal models fail to accurately replicate the anatomical features and operative logic of the human maxillary sinus [5]; commercial dental models solely focus on simulating hard tissues and cannot reproduce soft tissues such as the sinus mucosa and cysts; cadaver models are limited by high costs, ethical constraints, and resource scarcity, rendering them unfit for large-scale training [6]. This study specifically developed a customized 3D-printed model for medical education that can accurately replicate the complete clinical scenario of a maxillary sinus cavity, cyst, and large perforation [7]. The model comprehensively covers the entire surgical workflow, including lateral window maxillary sinus floor elevation, intra-sinus cyst removal, and repair of large perforations following cyst excision, providing physicians with a standardized, one-stop training platform. Materials and methods Study approval and participants Twenty postgraduate dental students from the School of Stomatology, Wuhan University (age range: 24–27 years), with basic clinical experience but no prior hands-on experience in this complex maxillary sinus perforation repair technique, were recruited as participants. All possessed two years of clinical practice experience (having participated in ≥ 30 implant surgeries) but no hands-on experience in maxillary sinus floor elevation and related complex procedures. All participants provided informed consent to participate in the study. The sample size of this study was determined with reference to previous preclinical simulation training studies in dental education [5, 12], which typically enrolled 15–25 participants to achieve sufficient statistical power for the Wilcoxon signed-rank test analysis of pre- and post-training outcomes. Fabrication of the 3D-printed training model for sinus floor elevation A clinical case meeting the following criteria was selected: a 23-year-old male with a missing right maxillary second premolar, first molar, and second molar; CBCT revealed residual sinus floor bone height < 3 mm, accompanied by an intra-sinus cyst measuring 8 mm in diameter. Data acquisition and processing Three-dimensional reconstruction of CBCT data in DICOM format was performed using InVesalius 3.0 software, with a focus on reproducing the following anatomical structures: stratified structure of maxillary cortical bone/cancellous bone, maxillary sinus cavity (volume approximately 15 mL), Schneiderian membrane (thickness 0.2 mm), and antral cyst (including cystic fluid-filled cavity). Selection of printing equipment and materials To precisely replicate clinical handling sensations, material selection adhered to the principle of anatomical function simulation, with the customized solution as follows: Maxilla: Changzhou Puliseng RP-405-TZ09G photopolymer resin (tensile strength 56–75 MPa, hardness D88, elastic modulus 2200–2600 MPa) was used to simulate the wear resistance of real cortical bone and ensure the grinding feel of bone windows. Maxilla: Changzhou Puliseng RP-405-TZ09G photopolymer resin (tensile strength 56–75 MPa, hardness D88, elastic modulus 2200–2600 MPa) was used to simulate the wear resistance of real cortical bone and ensure the grinding feel of bone windows. Schneiderian membrane: Medical-grade silicone rubber with Shore A hardness 30–40 (elongation at break 300%, tear strength 15 kN/m) was adopted to replicate the elasticity and resilience of the sinus floor, prevent tactile distortion, and foster gentle dissection techniques. Schneiderian membrane: Medical-grade silicone rubber with Shore A hardness 30–40 (elongation at break 300%, tear strength 15 kN/m) was adopted to replicate the elasticity and resilience of the sinus floor, prevent tactile distortion, and foster gentle dissection techniques. Maxillary sinus cyst: Designed as a detachable silicone capsule containing transparent gel (capsule wall thickness 0.1 mm, burst pressure 5 kPa) to simulate the feedback of cyst puncture, cyst fluid aspiration, and cyst wall removal. Maxillary sinus cyst: Designed as a detachable silicone capsule containing transparent gel (capsule wall thickness 0.1 mm, burst pressure 5 kPa) to simulate the feedback of cyst puncture, cyst fluid aspiration, and cyst wall removal. A Prismlab RP-400D-ACP-1 stereolithography 3D printer (printing precision 0.02 mm) was used with optimized parameters: layer thickness 100 μm; exposure time 8–10 s for cortical bone regions and 5–6 s for cancellous bone regions; and printing orientation aligned along the maxillary sinus’s longitudinal axis to avoid obstructing critical surgical areas. Post-printing standardized processing included ultrasonic cleaning with 99.5% isopropyl alcohol for 5 min (to remove residual resin), followed by UV irradiation at 365 nm for 20 min to achieve > 95% curing. A removable sinus cyst module was assembled to form the complete training model (Fig. 1). The model underwent CBCT scanning, with preoperative scan images retained (Fig. 2). Complete customized 3D-printed training model for large maxillary sinus membrane perforation repair Preoperative CBCT of the model: reconstructed maxillary sinus cavity, intra-sinus cyst Simulation of model surgery A self-controlled design was employed, with all participants receiving standardized training based on the 3D-printed models. This duration was selected based on previous preclinical implant simulation studies [5, 12], which validated that 2.5–3.5 h of structured training sufficiently covers theory, demonstration, hands-on practice, and feedback for complex maxillary sinus surgery while maintaining trainee concentration. The whole training was continuously supervised by associate chief physicians or above, with real-time correction of improper operations [5]. Preoperative evaluation and preparation A pre-training (pre-operative) baseline assessment was implemented for the simulated procedure, with the assessment results serving as a reference for subsequent skill evaluation and post-training outcome comparison. A 10-point Likert scale was employed to evaluate training effectiveness across five core dimensions, and the specific evaluation items are detailed in the supplementary questionnaire file. Scores ranged from 1 (completely unmastered) to 10 (completely mastered). Based on CBCT images from clinical cases, this training session first explained the core procedures of lateral window maxillary sinus elevation and key considerations for managing sinus cysts, with a focus on the formation mechanisms and repair challenges of large perforations following cyst removal. The training content included two key components: (1) The core principle for repairing large perforations is to completely isolate the maxillary sinus, preventing leakage of bone graft material, thereby effectively preventing infection and improving the success rate of bone grafting [9]; (2) Hands-on demonstration of specialized instrument usage (e.g., round drills, maxillary sinus membrane elevators, and membrane tack placement instruments) as well as key techniques for trimming and securing resorbable collagen membranes [10]. Simulated practice The training was carried out in strict accordance with clinical surgical protocols: Create a 15–20 mm × 8–10 mm elliptical bone window along the pre-marked lines, during which the bone window is retained as a complete bone piece and not attached to the soft tissue, ensuring smooth edges without serrated defects (Fig. 3a-b). The bone window pre-marked lines were uniformly set on all models before training, using a standardized 3D localization scheme tailored to the patient’s anatomy and clinical condition. The precise parameters were: inferior border 3 mm above the maxillary sinus floor, anterior border 3 mm distal to the second premolar root apex, posterior border 3 mm mesial to the sinus posterior wall, vertical height 8–10 mm, and mesial-distal length 15–20 mm. Create a 15–20 mm × 8–10 mm elliptical bone window along the pre-marked lines, during which the bone window is retained as a complete bone piece and not attached to the soft tissue, ensuring smooth edges without serrated defects (Fig. 3a-b). The bone window pre-marked lines were uniformly set on all models before training, using a standardized 3D localization scheme tailored to the patient’s anatomy and clinical condition. The precise parameters were: inferior border 3 mm above the maxillary sinus floor, anterior border 3 mm distal to the second premolar root apex, posterior border 3 mm mesial to the sinus posterior wall, vertical height 8–10 mm, and mesial-distal length 15–20 mm. Gently dissect the Schneiderian membrane until fully elevated, aspirate cyst fluid via puncture, and completely remove the cyst wall (Fig. 4a-b). Gently dissect the Schneiderian membrane until fully elevated, aspirate cyst fluid via puncture, and completely remove the cyst wall (Fig. 4a-b). Elevating the schneiderian membrane: continue to carefully elevate the schneiderian membrane with a dissector to form a bone augmentation space (Fig. 5). Elevating the schneiderian membrane: continue to carefully elevate the schneiderian membrane with a dissector to form a bone augmentation space (Fig. 5). Precisely trim the resorbable collagen membrane to standard dimensions of 30×40 mm. Secure its inner margin to the mesial and distal paranasal bone walls near the perforation (10 mm from the sinus floor) (Fig. 6a-b). Precisely trim the resorbable collagen membrane to standard dimensions of 30×40 mm. Secure its inner margin to the mesial and distal paranasal bone walls near the perforation (10 mm from the sinus floor) (Fig. 6a-b). Using a membrane stapler, vertically implant two 3 mm titanium staples into the bone wall (Fig. 7). Using a membrane stapler, vertically implant two 3 mm titanium staples into the bone wall (Fig. 7). Fill with simulated bone graft material matching Bio-Oss in composition and particle size, compacting it evenly. Graft placement was considered complete when the material filled the submembranous space uniformly without dead space and reached the planned augmentation height, and no extrusion occurred after gentle compression. The procedure was verified in real time by supervising physicians at the associate chief physician level or above. Fold the resorbable collagen membrane over the alveolar ridge to envelop the material, forming a sausage-like structure. Secure the lateral edge of the membrane with two additional 3 mm titanium nails on the lateral sinus wall or alveolar ridge, ensuring tight adhesion between the membrane, bone window edges, and material surface to prevent leakage (Fig. 8a-c). Fill with simulated bone graft material matching Bio-Oss in composition and particle size, compacting it evenly. Graft placement was considered complete when the material filled the submembranous space uniformly without dead space and reached the planned augmentation height, and no extrusion occurred after gentle compression. The procedure was verified in real time by supervising physicians at the associate chief physician level or above. Fold the resorbable collagen membrane over the alveolar ridge to envelop the material, forming a sausage-like structure. Secure the lateral edge of the membrane with two additional 3 mm titanium nails on the lateral sinus wall or alveolar ridge, ensuring tight adhesion between the membrane, bone window edges, and material surface to prevent leakage (Fig. 8a-c). Maxillary sinus lateral wall bone window preparation. a Marking of bone window position; (b) Bone window drilling with a 4.5 mm diamond ball bur Schneiderian membrane dissection and cyst removal. a Membrane dissection with a dissector; (b) Cyst fluid aspiration with a 5 ml syringe; (c) Perforation enlargement and residual cyst wall resection Schneiderian membrane elevation for bone augmentation space formation Resorbable collagen membrane preparation and placement. a Membrane trimming and folding; (b) Saline moistening and membrane placement under elevated membrane and on sinus lateral wall Collagen membrane fixation on maxillary sinus medial wall with a membrane tack placement instrument Bone graft filling and membrane final fixation. a Sinus floor bone graft filling; (b) Collagen membrane wrapping of bone graft material; (c) Membrane fixation on sinus lateral wall Review and reinforcement After the simulation exercise, the operated model was subjected to another CBCT scan to obtain simulated postoperative status images (Fig. 9). Students watched actual surgical videos and, combined with these postoperative scan images, compared and analyzed their own operational procedures against clinical standards to reinforce memory of the surgical workflow. After the training, a post-training questionnaire was distributed to evaluate the training effectiveness across the aforementioned five core dimensions. Postoperative CBCT of the model after sausage technique: bone graft position and perforation repair effect Data analysis Statistical analyses were conducted using IBM SPSS Statistics for Windows, Version 27.0 (IBM Corp., Armonk, NY, USA). Data normality was verified using the Shapiro-Wilk test. The Wilcoxon signed-rank test was employed to compare pre- and post-training score differences, with a significance level set at p < 0.05. Cronbach’s alpha coefficient was used to assess the reliability of the questionnaire. Cost analysis The full production cycle cost was calculated based on 2026 market prices, including: (1) Material costs: Resin 30 g/US$6.75, silicone rubber 5 g/US$0.90, auxiliary accessories US$0.90; (2) Equipment depreciation: The RP-400D printer unit price is $20,250, depreciated over 5 years, averaging 500 models printed annually; (3) Labor costs: Data processing + post-printing handling, $4.50 per model. Economic benefits were evaluated through cost comparisons with commercial dental implant models and cadaver models [6, 11]. Results Model evaluation All 20 participants completed the 3-hour standardized training and questionnaire completion, yielding 20 valid questionnaires. Cronbach’s alpha coefficients for the five core dimensions were 0.982 and 0.976 before and after training, respectively, indicating good measurement reliability (Table 1). The Shapiro-Wilk test revealed a non-normal distribution of the dataset, while the Wilcoxon signed-rank test demonstrated significant differences between pre- and post-training scores for all five dimensions (p < 0.001, Table 1). Table 1 Pre- and post-training questionnaire scoresEvaluation itemsBefore training (Mean ± SD)After training (Mean ± SD)p value (Wilcoxon test)Maxillary sinus anatomical understanding5.45 ± 1.578.60 ± 1.05<0.001Lateral approach anatomical localization5.00 ± 1.598.40 ± 1.05<0.001Lateral sinus floor elevation procedure mastery5.40 ± 1.318.45 ± 1.10<0.001Concurrent cyst enucleation key points comprehension4.25 ± 1.898.35 ± 1.09<0.001Large perforation repair technique mastery4.15 ± 1.738.20 ± 1.03<0.001Twenty clinicians completed 10-point Likert pre-/post-training questionnaires. Pre- and post-training Cronbach’s alpha were 0.982 and 0.976 (good reliability). Non-normal data (Shapiro-Wilk) showed significant post-training score improvements (Wilcoxon, all p<0.001) Pre- and post-training questionnaire scores Twenty clinicians completed 10-point Likert pre-/post-training questionnaires. Pre- and post-training Cronbach’s alpha were 0.982 and 0.976 (good reliability). Non-normal data (Shapiro-Wilk) showed significant post-training score improvements (Wilcoxon, all p<0.001) All participants acknowledged the training model’s value: it clearly depicted the three-dimensional anatomical relationships between the maxillary sinus, cysts, and perforations; its operational feel closely resembled clinical practice [7]; it effectively aided understanding of key steps in repairing large perforations [8, 10]; and it significantly enhanced practical proficiency. Cost evaluation After calculation, the unit cost of the 3D-printed model comprised material costs of $8.55, equipment depreciation of $1.35, and labor costs of $5.85, resulting in a single-session training cost of $15.75. Compared to traditional methods, the advantages were remarkably significant: the single-use training cost for commercial dental implant models was $112.50 (based on standardized market pricing for commercial dental training models and clinical training cost statistics), while the cost of the 3D-printed models in this study was only 14% of that amount; compared to cadaver models costing $67.50 per training session (calculated based on actual anatomical material consumption, referencing medical education specimen cost standards, including specimen procurement, preservation, and disposal fees), the cost of this 3D-printed model was only 23.3% of that amount [6, 11]. For a 20-person training session, the total cost was $315. This represented a savings of $1,935 compared to commercial dental implant models (total cost: $2,250) and $1,035 compared to cadaver models (total cost: $1,350), demonstrating outstanding economic viability [6, 11]. Discussion In clinical practice, repairing large perforations during lateral window maxillary sinus elevation combined with cyst enucleation is a highly challenging procedure. Clinicians with limited experience in maxillary sinus surgery, lacking specialized hands-on training, often struggle to accurately navigate anatomical structures and control surgical force, leading to repair failure and subsequent complications [4, 11]. The core challenge in large perforation repair lies in clinical execution [7]. Traditional theoretical instruction relies solely on two-dimensional images to convey procedural essentials, hindering the development of three-dimensional spatial cognition. The 3D-printed model, however, physically represents anatomical structures; combined with the visualization advantages of CBCT scans, it enables practitioners to intuitively observe the relationship between the perforation and surrounding tissues. This facilitates precise mastery of critical techniques—such as the positioning and tension required for resorbable collagen membrane fixation [9]—significantly enhancing procedural accuracy [10]. In dental training, access to high-quality hands-on models is a critical factor limiting technology adoption [12]. Cadaver models are costly and subject to ethical constraints, while commercial models lack customization and soft tissue simulation capabilities, failing to meet specialized training needs [6, 11]. The 3D-printed model developed in this study replicates clinical scenarios at a 1:1 scale to faithfully recreate the full workflow of bone window preparation, cyst removal, and large perforation repair [7]; it not only provides clinicians with a zero-risk hands-on training platform to effectively shorten their learning curve [5], but also features low-cost materials and a short production cycle, with anatomical parameters adjustable according to CBCT data from different clinical cases [7]. This adaptability makes the model well-suited for diverse training scenarios and thus offers a feasible solution for the standardized training of complex perforation repair techniques [12]. Furthermore, this customized model enables preoperative rehearsal for complex maxillary sinus perforation repair, which can improve surgical accuracy and success rate, reduce risks, and strengthen its clinical value. This study has certain limitations: First, differences exist between the simulated materials and native human tissues. The elastic modulus of the silicone rubber used to simulate the sinus mucosa is slightly higher than that of native maxillary sinus mucosa, and the resin material simulating bone is more brittle, which may slightly affect the tactile feedback during surgery. Future work will develop bionic silicone materials with elastic modulus closer to native maxillary sinus mucosa, as well as gradient composite resins that better mimic the structure and texture of natural bone. Second, long-term clinical follow-up was not conducted. The study only evaluated immediate training outcomes, failing to fully validate the correlation between model-based training and clinical repair success rates [4, 9]. Subsequent tracking of participants’ clinical performance is needed to further validate the model’s long-term educational value [11, 12]. In future studies, we will develop an intraoral simulation model to replicate real surgical vision and operator position, record total operation time, and analyze the learning curve to further validate training efficacy. With the deep integration of 3D printing and artificial intelligence, this model can be upgraded into an intelligent training system by integrating AI image recognition to automatically compare operations with standard procedures and generate personalized feedback [7]. It can also be extended to other complex oral and maxillofacial surgeries [12]. Conclusion Personalized 3D-printed models offer an efficient, clinically applicable and cost-effective training solution for large maxillary sinus membrane perforation repair. The models accurately replicate the complete clinical scenario of large perforations following lateral window maxillary sinus floor elevation combined with cyst enucleation. Through standardized hands-on training, they significantly improve anatomical cognition, operational proficiency and clinical confidence of clinicians with limited maxillary sinus surgery experience, effectively shortening the learning curve of this complex procedure and reducing surgical risks. With high simulation fidelity, customizable design and low cost, the model addresses core pain points of traditional training tools including inaccurate scenario simulation, high cost and ethical constraints. It provides an innovative tool for standardizing the promotion of complex perforation repair techniques in dentistry, helping to elevate technical proficiency in primary healthcare institutions and improve overall outcomes of maxillary sinus elevation procedures. Supplementary Information Supplementary Material 1. Supplementary Material 2. Supplementary Material 3. Acknowledgements The authors would like to appreciate all participating clinicians for their cooperation in the study. Authors’ contributions H.L. designed the study, fabricated the 3D model, performed experiments, and analyzed data.Y.S. curated data, drafted the original manuscript, and edited the final version.Y.C., M.Z., and S.X. collected data, supervised experimental procedures, and managed quality control.Y.Sa. conceptualized the work, revised the manuscript, supervised the study, and administered the project.All authors reviewed the manuscript. Funding This work was supported by the Fundamental Research Funds for the Central Universities (Grant No. 2042025YXB022). Data availability The data and materials supporting the conclusions of this study are available from the corresponding author upon reasonable request. Declarations Ethics approval and consent to participate This study was conducted in full compliance with the Declaration of Helsinki (2024 version) and approved by the Medical Ethics Committee of Yaboshi Dental (Ethical Approval No. [2025] EC20191203-1005). The ethical approval is valid for 24 months. All participants were fully informed of the purpose, procedures, and potential benefits of the study and provided written informed consent prior to participation. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Haoxin Lv and Yitong Sun contributed equally to this work. References Associated Data Supplementary Materials Supplementary Material 1. Supplementary Material 2. Supplementary Material 3. Data Availability Statement The data and materials supporting the conclusions of this study are available from the corresponding author upon reasonable request.