This single-center prospective study was conducted at a tertiary care teaching hospital in India. A total of 30 plastic surgery residents were enrolled following an open call for voluntary participation. All participants provided written informed consent prior to the study. Since the research involved the use of anonymized CT imaging data and no identifiable patient information or clinical intervention, the ethical approval was waivered. A waiver was taken from Institutional Ethical Committee (IEC) for this study.

The primary objective of the study was to evaluate surgical skill development in two technically demanding tasks relevant to head and neck reconstruction: (1) Fibula osteotomy planning and execution and (2) soft tissue flap inset on 3D-printed models with bony and soft tissue defect. The study was simulation based and did not involve any live surgical procedures.

CT scan images of a de-identified patient with a segmental mandibular defect were used to generate three-dimensional models. The images were converted from DICOM format to STL files using 3D Slicer, an open-source medical image segmentation software. Post-processing of the STL files was carried out using Blender to remove unwanted anatomical segments and Meshmixer to eliminate artifacts and refine the surface geometry. The finalized STL model was converted into a g-code format using Cura slicing software, enabling fabrication with a fused deposition modeling (FDM) 3D printer.

The in-house printer used had a build volume of 190 × 190 × 160 mm and nozzle diameter of 0.4 mm and supported PLA, PETG, and TPU filaments. Layer height ranged between 0.1 and 0.2 mm. Printing was carried out using PLA filament (1.75 mm) sourced from Solidspace Technology LLP, India. The printed mandible and maxilla were complemented by a simulated soft tissue component—namely, a tongue cast from platinum-grade silicone rubber using a 3D-printed mold. A mucosal lining was created using foam sheets, and all components were assembled with silicone-based adhesive. The silicone rubber was procured from Chemzest Techno products Pvt. Ltd. India. The cost per model was ₹1400 (~ US $16.39), with a breakdown as follows: PLA (150 g) ₹230, silicone (550 g) ₹900, silicone adhesive ₹100, and electricity used during 3D printing ₹170 (Fig. 1).

Composite image showing 3D-printed training models. A Marginal mandibulectomy with upper alveolectomy and resected buccal mucosa represented by blue foam. B Hemiglossectomy defect. The resected tongue is shown in brown, and surrounding buccal and pharyngeal mucosa are depicted using blue foam. C Posterior segmental mandibulectomy involving the posterior mandibular segment and buccal mucosa represented by blue foam

A 3D template of the fibula was derived from patient leg CT data and replicated using acrylic sheets. These sheets were manually cut into fibula shapes on a tilting table with a motorized cutting saw. The final construct had anatomical accuracy and similar density to bone, allowing for realistic osteotomy and plate fixation. The model incorporated several features as follows: Red foam to represent the flexor hallucis longus muscle, copper wire to mimic vascular perforators, yellow foam to simulate the skin paddle, and clear tape to replicate the interosseous septum. Each model cost ₹100 (~ US $1.17), with an additional ₹200 (~ US $2.34) spent on supporting stationery items (Fig. 2).

Composite image showing custom-made acrylic fibula model. A Acrylic white fibula segment with red foam representing the flexor hallucis longus (FHL), red copper wire mimicking vascular perforators, and yellow foam with transparent cellotape representing the skin paddle and septum. Blue and red copper wires represent the main pedicle. B Oblique view of the same model highlighting the anatomical relationship between fibula, muscle, perforator vessels, and septocutaneous paddle, designed to aid osteotomy and inset training

The training was delivered in two phases: an introductory lecture followed by two hands-on simulation modules. The first session focused on osteotomy planning and execution using the acrylic fibula constructs. The second involved flap inset onto the 3D-printed mandibular models, which included silicone-based soft tissues. Participants assembled their models using 1.5-mm titanium miniplates and screws and used standard surgical drills, burrs, and instruments under faculty supervision (Figs. 3 and 4).

Composite image showing osteotomy planning and execution. A Custom-made acrylic fibula model before planning osteotomy. B Planning of osteotomy on the right side and fibula model after performing osteotomy on left

Image showing soft tissue inset of hemiglossectomy defect. Inset is represented by red foam

Participant performance was assessed using the Zwisch scale, a validated four-level scale developed to evaluate surgical autonomy in the context of competency-based education [9]. The scale includes Level 1 (“show and tell”), where the instructor performs and narrates the procedure; Level 2 (“active help”), where the trainee performs parts of the procedure with continuous instruction; Level 3 (“passive help”), where the trainee performs most steps independently with occasional input; and Level 4 (“supervision only”), where the trainee operates independently with the supervisor observing silently. Two blinded senior consultants independently evaluated each participant’s performance in three domains: fibula osteotomy execution, flap inset, and instrument handling. Discrepancies in scoring were resolved by consensus.

In addition to objective evaluation, participants completed a six-item questionnaire (Table 1) based on a 4-point Likert scale to assess self-perceived improvement in anatomical understanding, surgical technique, instrument use, and operative planning. This questionnaire format has been used in previous simulation-based education studies and includes statements ranging from “I was able to understand the contents of the lecture” to “I was able to understand the importance of surgical planning.”

All data were analyzed using IBM SPSS Statistics version 25.0. Pre- and post-training questionnaire scores were treated as paired ordinal data and compared using the Wilcoxon signed-rank test. Results were expressed as mean ± standard deviation, with p-values < 0.05 considered statistically significant.