Evaluating the Efficacy of a Novel Multimaterial 3D-Printed Phantom for Otologic Surgical Education and Simulation

Abstract

Objective:

To design and validate a cost-effective, multimaterial 3D-printed temporal bone phantom that replicates the anatomic and haptic properties of human bone for surgical simulation and training.

Study design:

Prospective cohort study.

Setting:

Tertiary academic medical center.

Participants:

Sixteen otolaryngology residents (PGY-1 to PGY-5) are participating in a multi-institutional temporal bone dissection course.

Intervention:

Participants performed mastoidectomy and facial recess dissection on a novel 3D-printed phantom fabricated from calcium-enriched SimuBone PLA (for osseous structures) and contrasting colored PLA (for neurovascular anatomy). Performance was compared against cadaveric controls.

Main outcome measure(s):

Participants completed a structured 5-point Likert-scale survey assessing anatomic realism, tactile feedback, visual fidelity, and educational utility. Spearman rank-order correlations evaluated the relationship between surgeon experience and perceived model fidelity.

Results:

The phantom demonstrated moderately high scores for anatomic fidelity (mean 3.56 ± 0.51) and visual realism (mean 3.62 ± 0.50), with overall realism rated at 3.50 ± 0.52. Although tactile realism was rated lower (mean 3.06 ± 0.57), 100% of participants rated the model as moderately to highly valuable for surgical training. Senior residents were more critical of anatomic fidelity (P < 0.05) yet reported continued willingness to use the phantom for skill development. Qualitative feedback confirmed the utility of color-coded “no-go” zones for novice spatial orientation.

Conclusions:

The multimaterial 3D-printed phantom offers a scalable, cost-effective ($6.13 USD) alternative to cadaveric dissection with high anatomic accuracy. Although haptic nuances require further optimization, the model serves as an effective foundational simulator for otologic surgical education.

Introduction

Surgical competency in otology and neurotology requires a robust understanding of temporal bone anatomy and the ability to navigate the intricate relationships among the facial nerve, labyrinth, and major vascular structures.^[1–3] These psychomotor and spatial skills have traditionally been acquired through repetitive cadaveric dissection, which remains the gold standard for training.^[4,5] However, the utility of cadaveric specimens is increasingly limited by high procurement costs, limited availability, and the biohazardous nature of tissue preservation.^[6–8] Furthermore, the variability in pneumatization and pathology among cadaveric specimens can make standardized training difficult to implement across residency programs.^[9]

In response, several alternative training modalities have been explored, including virtual reality simulators, synthetic models, and 3D-printed phantoms.^[10–12] While virtual platforms offer repeatability and objective performance tracking, many cannot fully replicate the tactile and tissue responses of real bone, including the differences in density, drill resistance, and the production of bone dust or debris.^[13,14] Early synthetic and single-material models likewise failed to capture the heterogeneous microstructure of the temporal bone, limiting their effectiveness in teaching residents to distinguish dense otic capsule from aerated mastoid bone.^[15,16]

More sophisticated resin-based stereolithography and digital light processing models achieve high spatial resolution but tend to be brittle and produce sticky or toxic particulates during drilling.^[17–19] Fused-deposition models fabricated from standard PLA offer cost-effectiveness and accessibility but often suffer from low thermal stability, causing material to melt or smear rather than generate bone-like chips.^[20,21] Ceramic-based or composite models reproduce granular bone removal but are frequently fragile, expensive, or restricted to monochromatic printing, limiting visual identification of critical neurovascular structures.^[22,23]

There remains a gap in the availability of cost-effective training models that bridge the divide between anatomic accuracy and haptic fidelity.^[24] Recent advances in multimaterial 3D printing offer a promising solution by enabling the fabrication of anatomically precise, multidensity phantoms that can mimic both the bony and soft-tissue components of the temporal bone.^[25] Unlike single-substrate additive manufacturing, multi-material printing allows simultaneous deposition of distinct polymers with differing mechanical properties.^[26,27] This approach facilitates the creation of phantoms that replicate the specific drilling characteristics of human bone while visually differentiating soft-tissue structures such as the dura mater, sigmoid sinus, and facial nerve.^[28–30]

The objective of the present study is to evaluate the efficacy of a novel multimaterial 3D-printed temporal bone phantom designed for otologic surgical training. This investigation assesses the anatomic accuracy, tactile realism, and educational utility of the phantom through a comparative analysis involving otolaryngology residents and senior surgeons. By validating the performance of this phantom against the established standards of cadaveric dissection, this research aims to determine the viability of multimaterial printing as a scalable and effective tool for modern otologic surgical education.

Materials and methods

Phantom design and fabrication

Image acquisition and segmentation

A deidentified, high-resolution cadaveric temporal bone dynamic computed-tomography (DynaCT) scan served as the source data for the phantom (Fig. 1). DICOM data were imported into 3D Slicer for preprocessing. Semiautomated thresholding combined with manual segmentation delineated osseous structures, including the mastoid bone, bony labyrinth, vestibular aqueduct, ossicular chain, and internal auditory canal. Key soft-tissue and neurovascular landmarks, including the facial nerve, chorda tympani, internal carotid artery, sigmoid sinus, dura mater, and vestibulocochlear nerve branches, were also isolated.

Figure 1.

Preparation and fabrication workflow for the multimaterial 3D-printed temporal bone phantom. (Left) Segmentation: a high-resolution dynamic CT (DynaCT) scan was processed in 3D Slicer to isolate the osseous anatomy and outline critical neurovascular structures. (Center) Computer-aided design (CAD): the segmented models were imported into Blender for mesh refinement and the integration of lateral support blocks, designed to facilitate bench-top stabilization and fiducial marker placement. The supports are shown here in yellow for clarity, but in the printed phantom, they are colored to match the adjacent dura mater and sigmoid sinus. (Right) Multimaterial printing: the final assembly was fabricated on a Prusa XL Original 3D printer, utilizing distinct filaments and colors for the bone, soft tissue, and support casing.

Computer-aided design and model preparation

The segmented masks were exported as Stereolithography (STL) files and imported into Blender (Blender Foundation, Amsterdam, The Netherlands) for mesh refinement and assembly. Two lateral mounting blocks were designed and integrated into the external surface of the model to facilitate bench-top stabilization and potential registration for image-guided surgery. The temporal bone mesh was designated as the primary reference object to preserve spatial fidelity among internal anatomic structures. All neurovascular and labyrinthine components were subsequently aligned to this reference before final export as individual STL files.

Slicing and print configuration

The component STL files were imported into the PrusaSlicer software for fabrication preparation. The temporal bone model was loaded first, followed by soft-tissue and support elements fixed to the primary coordinate system. Print parameters were optimized to balance anatomic detail with fabrication time, employing a 0.10 mm layer height (“FAST DETAIL”) and a 0.4 mm nozzle. Variable infill settings, determined through preliminary drill testing, were applied to approximate human bone density while preserving the integrity of the simulated air cells. The temporal bone component was printed at 50% infill to simulate cortical hardness, whereas peripheral structures used 15% infill. Support material generation was managed using the “Organic” style with a 40-degree overhang threshold and a branch density of 30% to facilitate manual removal post-production without damaging delicate surface features. To prevent unintentional adhesion to bone-simulating material and reduce overall printing cost, all support structures were printed using standard PLA filament rather than the specialized bone-mimicking material.

Multimaterial fabrication

A Prusa XL Original (Prusa Research, Prague, Czech Republic) 3D printer with a 5-toolhead system was used to fabricate the phantom. For this study, the printer was configured to operate using 3 toolheads, corresponding to the 3 distinct material groups: osseous structures, critical soft-tissue structures, and the external support casing with dura and sigmoid sinus (Fig. 2). Osseous components were produced using SimuBone (3DXTech, Grand Rapids, Michigan) PLA at 1.75 mm diameter, a calcium-enriched filament engineered to better replicate the tactile feedback of human bone, providing greater resistance, reduced thermal softening, and improved chip formation compared with conventional PLA.^[31,32] Critical soft-tissue structures were printed in contrasting colored Prusament PLA (Red) to provide clear visual landmarks and delineate surgical “no-go” zones during dissection. The external support casing, as well as the sigmoid sinus and dura mater, were printed in a third color of Prusament PLA (Blue). Material assignment was managed through the slicer’s multiple-extruder configuration to ensure consistency across components.

Figure 2.

Component breakdown and material composition of the multimaterial 3D-printed temporal bone phantom. A, Fully assembled phantom illustrating the integration of osseous, soft tissue, and support structures. B, The osseous temporal bone segment is fabricated from SimuBone PLA to mimic the tactile feedback of drilling human cortical bone. C, The support casing and venous structures (ie, dura mater and sigmoid sinus) are printed in blue Prusament PLA to enable bench-top stabilization and delineate surgical boundaries. D, Critical neurovascular and conductive anatomy (ie, internal carotid artery, facial nerve, ossicular chain, etc.) are printed in contrasting red Prusament PLA to serve as clear visual landmarks and “no-go” zones during dissection.

Multi-institutional temporal bone course evaluation

Course design and participants

The temporal bone course was conducted at a single institution and included residents from 3 academic, nonprofit training programs to broaden the diversity of surgical backgrounds within the cohort. The course was structured over 2 separate days to accommodate varying levels of surgical experience: the first day was designated for junior residents (PGY1–3), and the second day for senior residents (PGY3–5). The study was conducted under IRB approval (IRB00264318), and all participants provided oral informed consent before participation. Training began with hands-on drilling of the multimaterial 3D-printed temporal bone phantom. Residents were oriented to the model’s anatomic landmarks, tactile feedback, and spatial relationships. Those with prior experience or demonstrated proficiency were guided through a mastoidectomy with identification of the facial recess, allowing for progressive skill development in a controlled environment. Following phantom-based training, participants performed cadaveric dissection on one ear of a whole human head cadaver, consolidating the techniques and anatomic knowledge gained from the synthetic model.

Survey instrument and data collection

Participants completed a structured, anonymous survey following hands-on interaction with both the multimaterial phantom and cadaveric temporal bones, allowing them to directly compare the fidelity and educational utility of the phantom relative to human temporal bone. The survey collected demographic and training information, including PGY level, prior mastoidectomy experience, and self-reported familiarity and proficiency with temporal bone drilling. Participants then rated key aspects of the phantom using a 5-point Likert-scale, evaluating both the realism of each domain and its perceived importance for surgical training. Domains included anatomic appearance, fidelity of internal structures, tactile feedback during drilling, and overall realism compared with the human temporal bone. Realism was scored from 1 (“not at all realistic”) to 5 (“nearly identical to human anatomy”), whereas importance was scored from 1 (“not at all important”) to 5 (“essential”). Additional items assessed perceived value for surgical training, likelihood and frequency of future use and open-ended feedback for model improvement.

Statistical analysis

Survey data were analyzed using Python (3.13). All Likert-scale realism and importance ratings were treated as ordinal variables. PGY level, prior mastoidectomy volume, and self-rated drilling familiarity were encoded as ordered categories. Prior mastoidectomy volume was further mapped to approximate numeric midpoints (0; 1–5 → 3; 6–10 → 8; 11–20 → 15; 21–50 → 35) to allow nonparametric correlation analysis. Associations between realism ratings and resident experience were evaluated using Spearman rank-order correlations, with the resulting correlation coefficient (ρ) serving as the primary measure of effect size. Three experience indicators were examined independently: (1) PGY level, (2) estimated number of prior cortical mastoidectomies, and (3) self-reported familiarity with temporal bone drilling.

Due to the pilot nature of the study and the limited sample size (n = 16), multivariable adjustment for collinear confounders (eg, PGY level and prior surgical volume) was not feasible. To avoid inflating type II error in this exploratory context, formal adjustments for multiple comparisons were also not applied. For aggregate survey results and correlation analyses, 95% CIs are reported alongside standard P-values, with statistical significance defined as P < 0.05 for all analyses.

Results

Cost and time of phantom fabrication

The cost and time required for fabricating a single multimaterial temporal bone phantom were precisely tracked to ensure reproducibility and assess scalability. Utilizing the stated configuration on the Prusa XL Original 3D printer, the average print time for a single phantom was ~12.25 hours. Postprocessing, primarily consisting of support removal and minor surface cleanup, required an additional 10 minutes per model. The total material cost per phantom was calculated to be $6.13 USD, demonstrating a cost-effective alternative to cadaveric specimens (Table 1). The specialized bone-analog filament (SimuBone PLA) accounted for most of this expense ($4.31), while the standard PLA used for soft-tissue landmarks and structural support contributed minimally to the overall cost. The fixed capital cost for the 3D printing hardware is ~$3,500 USD. Despite this initial investment, the low marginal cost per unit supports the financial viability of this model for high-volume surgical training programs.

Table 1.

Material cost breakdown for a single multimaterial 3D-printed temporal bone phantom

Material	Cost Per Spool (USD)	Total Grams Per Spool	Grams Used Per Phantom	Cost Per Phantom (USD)
SimuBone (1.75 mm)	$98.00	750	32.97	$4.31
PLA 1 (dura+sigmoid sinus+support)	$32.99	1,000	49.23	$1.62
PLA 2 (critical nerve+vascular structures)	$32.99	1,000	6.14	$0.20
Total material cost				$6.13

Itemized costs for materials used in the production of one multimaterial temporal bone phantom. Costs are calculated based on the price per spool, total filament mass per spool, and grams of each material required per phantom. SimuBone comprised the sole osseous substrate, while PLA materials were used to print dural structures, the sigmoid sinus, support material, and critical neurovascular elements. Total material cost reflects the sum of all components used to fabricate a single phantom.

Participant demographics

A total of 16 otolaryngology residents participated in the evaluation (Table 2). The cohort was predominantly female (81.25%) and universally right-hand dominant (100%). Participants represented a comprehensive range of training levels, with equal representation from junior (PGY-1 to PGY-2, n = 6) and senior (PGY-3 to PGY-5, n = 10) residents. Prior surgical experience varied appropriately with training level. 37.5% of participants (n = 6) had performed 5 or fewer cortical mastoidectomies, while 37.5% (n = 6) had performed more than 10. All participants self-rated their proficiency in temporal bone drilling as “somewhat proficient” (100%), and the majority described their familiarity with the procedure as “moderately familiar” (62.5%).

Table 2.

Participant demographics

Category	Subcategory	Count (%)
Sex	Male	3 (18.75)
	Female	13 (81.25)
Dominant hand	Right	16 (100.00)
	Left	0 (0.00)
Year in training	PGY-1	1 (6.25)
	PGY-2	5 (31.25)
	PGY-3	5 (31.25)
	PGY-4	3 (18.75)
	PGY-5	2 (12.50)
Familiarity with drilling temporal bone	Not at all familiar	0 (0.00)
	Somewhat familiar	6 (37.50)
	Moderately familiar	10 (62.50)
	Very familiar	0 (0.00)
Proficiency with drilling temporal bone	Not at all proficient	0 (0.00)
	Somewhat proficient	16 (100.00)
	Moderately proficient	0 (0.00)
	Very proficient	0 (0.00)
No. prior cortical mastoidectomies performed	0	2 (12.50)
	1–5	4 (25.00)
	6–10	4 (25.00)
	11–20	5 (31.25)
	21–50	1 (6.25)

Demographic and training characteristics of the study cohort (N = 16). Variables include sex, hand dominance, postgraduate year (PGY) level, self-rated familiarity and proficiency with temporal bone drilling, and the number of prior cortical mastoidectomies performed. Values are presented as counts with corresponding percentages.

Multimaterial 3D-printed phantom evaluation

Anatomic, visual, and haptic fidelity

Quantitative assessment of the phantom’s realism across 4 domains is summarized in Table 3. Anatomic accuracy received a mean rating of 3.56 (95% CI: 3.29–3.83) ± 0.51. Junior residents (PGY-2 and PGY-3) tended to assign higher ratings (means: 3.80 and 3.60) than more senior residents (PGY-4 and PGY-5; means: 3.33 and 3.00). Participants identified anatomic realism as the highest-priority feature, with an importance rating of 4.56 (95% CI: 4.22–4.90) ± 0.63. The macroscopic appearance of the model also received consistently favorable evaluations across all subgroups (mean: 3.62 [95% CI: 3.35–3.89] ± 0.50), with minimal variation across PGY levels (range: 3.33 to 4.00) or prior surgical experience (range: 3.00 to 4.00).

Table 3.

Realism ratings of the multimaterial 3D-printed temporal bone phantom

Domain	Subcategory	Mean±SD		95% CI
A. Aggregate Results (n=16)
Macroscopic appearance	Realism rating	3.62±0.50		[3.35–3.89]
	Importance rating	3.62±0.81		[3.19–4.05]
Anatomic accuracy	Realism rating	3.56±0.51		[3.29–3.83]
	Importance rating	4.56±0.63		[4.22–4.90]
Tactile feel	Realism rating	3.06±0.57		[2.76–3.36]
	Importance rating	4.12±0.72		[3.74–4.50]
Overall realism		3.50±0.52		[3.22–3.78]
PGY level	Appearance	Anatomy	Tactile feel	Overall
B. Stratified by PGY
PGY-1 (n=1)	4.00a	4.00a	4.00a	4.00a
PGY-2 (n=5)	3.60±0.55	3.80±0.45	3.20±0.45	3.60±0.55
PGY-3 (n=5)	3.80±0.45	3.60±0.55	3.00±0.00	3.80±0.45
PGY-4 (n=3)	3.33±0.58	3.33±0.58	2.67±0.58	3.00±0.00
PGY-5 (n=2)	3.50±0.71	3.00±0.00	3.00±1.41	3.00±0.00
Prior mastoidectomy cases	Appearance	Anatomy	Tactile feel	Overall
C. Stratified by number of prior cortical mastoidectomies performed
0 (n=2)	4.00±0.00	4.00±0.00	4.00±0.00	4.00±0.00
1–5 (n=4)	4.00±0.00	4.00±0.00	3.00±0.00	4.00±0.00
6–10 (n=4)	3.25±0.50	3.25±0.50	3.00±0.82	3.00±0.00
11–20 (n=5)	3.60±0.55	3.40±0.55	2.80±0.45	3.40±0.55
21–50 (n=1)	3.00a	3.00a	3.00a	3.00a
Self-rated drilling familiarity	Appearance	Anatomy	Feel	Overall
D. Stratified by self-rated familiarity with drilling the temporal bone
Somewhat familiar (n=6)	3.83±0.4 1	3.67±0.52	3.17±0.75	3.67±0.52
Moderately familiar (n=10)	3.50±0.53	3.50±0.53	3.00±0.47	3.40±0.52

Aggregate and stratified Likert-scale responses evaluating the realism of the simulated temporal bone phantom. Realism domains include appearance, anatomic accuracy, tactile feedback, and overall realism. Ratings were scored on a 5-point Likert-scale (1 = not at all realistic, 2 = somewhat realistic, 3 = moderately realistic, 4 = very realistic, 5 = nearly identical to human bone). Data are reported as mean±SD. Stratified analyses are presented by postgraduate year (PGY), number of prior cortical mastoidectomies performed, and self-rated familiarity with drilling the temporal bone. aSingle respondent, SD not applicable

Tactile feedback was rated lower than visual or anatomic fidelity (mean: 3.06 [95% CI: 2.76–3.36] ± 0.57), reflecting the subjective difference between the SimuBone material and native human bone. Residents with greater surgical experience (11 to 20 prior cases) tended to assign lower haptic realism scores (mean: 2.80) compared with novices with no prior cases (mean: 4.00). Despite this, the perceived importance of tactile realism remained high at 4.12 (95% CI: 3.74–4.50) ± 0.72, identifying it as a key target for future material optimization. Overall realism, defined as the global assessment of the phantom relative to the human temporal bone, achieved a mean score of 3.50 (95% CI: 3.22–3.78) ± 0.52.

Correlation between experience and realism ratings

Training level (PGY) demonstrated a significant negative correlation with anatomic realism (ρ = −0.552, [95% CI: −0.82 to −0.08], P = 0.027) and overall realism (ρ = −0.533, 95% CI: −0.81 to −0.05], P = 0.034), indicating that more senior residents tended to rate the phantom as less realistic. Prior mastoidectomy case volume showed similar patterns, with significant negative correlations for anatomic realism (ρ = −0.592, [95% CI: −0.84 to −0.14], P = 0.016) and overall realism (ρ = −0.587, [95% CI: −0.84 to −0.13], P = 0.017). Self-rated drilling familiarity showed no significant correlation with any realism domain (all P > 0.20), suggesting that objective prior operative exposure, rather than subjective confidence, influenced perception of model fidelity.

Educational value and likelihood of usage

Participants affirmed the educational utility of the multimaterial phantom, with 100% of respondents rating it as “moderately valuable” or higher (Table 4). The majority of the cohort (75%, n = 12) classified the model as “very valuable” or “essential” for surgical training. Stratification by surgical volume revealed that novices (0 to 5 prior cortical mastoidectomies) rated the educational value highest, with 100% (n = 6) selecting “very valuable” or “essential.” In contrast, residents with intermediate operative experience (6 to 20 cases) offered more varied assessments, with 44.4% (n = 4) rating the model as “moderately valuable.” A similar trend was observed across training levels: while junior residents (PGY-1 through PGY-3) assigned “essential” ratings (18.2%, n = 2), PGY-4 and PGY-5 residents rated the model exclusively as “moderately” or “very” valuable.

Table 4.

Perceived educational value and likelihood of use of the multimaterial 3D-printed temporal bone phantom

Category	Moderately Valuable (%)	Very Valuable (%)	Essential (%)
A. How valuable is the simulated temporal bone for surgical training? Only response categories endorsed by participants are shown. No participants selected “not at all valuable” or “somewhat valuable”
Overall (n=16)	4 (25.0)	10 (62.5)	2 (12.5)
By PGY level
PGY-1 (n=1)	0	1 (100.0)	0
PGY-2 (n=5)	1 (20.0)	3 (60.0)	1 (20.0)
PGY-3 (n=5)	1 (20.0)	3 (60.0)	1 (20.0)
PGY-4 (n=3)	1 (33.3)	2 (66.7)	0
PGY-5 (n=2)	1 (50.0)	1 (50.0)	0
By no. prior cortical mastoidectomies
0 (n=2)	0	2 (100.0)	0)
1–5 (n=4)	0	3 (75.0)	1 (25.0)
6–10 (n=4)	2 (50.0)	2 (50.0)	0
11–20 (n=5)	2 (40.0)	2 (40.0)	1 (20.0)
21–50 (n=1)	0	1 (100.0)	0
By self-rated familiarity with drilling the temporal bone
Somewhat familiar (n=6)	1 (16.7)	4 (66.7)	1 (16.7)
Moderately familiar (n=10)	3 (30.0)	6 (60.0)	1 (10.0)
Category	Neutral	Likely	Very Likely
B. If the phantom were readily available, how likely would you use it? Only response categories endorsed by participants are shown. No participants selected “very unlikely” or “unlikely”
Overall (n=16)	1 (6.2)	11 (68.8)	4 (25.0)
By PGY Level
PGY-1 (n=1)	0	0	1 (100.0)
PGY-2 (n=5)	0	4 (80.0)	1 (20.0)
PGY-3 (n=5)	0	4 (80.0)	1 (20.0)
PGY-4 (n=3)	1 (33.3)	1 (33.3)	1 (33.3)
PGY-5 (n=2)	0	2 (100.0)	0
By Number of Prior Cortical Mastoidectomies
0 (n=2)	0	1 (50.0)	1 (50.0)
1–5 (n=4)	0	4 (100.0)	0
6–10 (n=4)	1 (25.0)	2 (50.0)	1 (25.0)
11–20 (n=5)	0	3 (60.0)	2 (40.0)
21–50 (n=1)	0	1 (100.0)	0
By self-rated familiarity with drilling the temporal bone
Somewhat familiar (n=6)	0	4 (66.7)	2 (33.3)
Moderately familiar (n=10)	1 (10.0)	7 (70.0)	2 (20.0)

Responses to survey items assessing the perceived educational value of the multimaterial 3D-printed temporal bone phantom (Part A) and the likelihood of use if the phantom were readily available (Part B). Only response categories endorsed by participants are reported for clarity. No respondents selected “not at all valuable” or “somewhat valuable” in Part A, nor “very unlikely” or “unlikely” in Part B. Values are presented as counts with corresponding percentages and are shown overall and stratified by postgraduate year (PGY), number of prior cortical mastoidectomies performed, and selfrated familiarity with drilling the temporal bone.

Regarding potential adoption into the residency curriculum, 93.8% (n = 15) of participants indicated they would be “likely” or “very likely” to use the phantom if it were readily available. The single “neutral” response was recorded by a PGY-4 resident. Notably, despite the lower realism scores assigned by PGY-5 residents in anatomic and haptic domains, 100% of this subgroup indicated they would be “likely” to utilize the model, suggesting that the phantom’s perceived utility for training persists independent of its limitations in absolute fidelity. Among participants who indicated a likelihood of adopting the phantom into their practice (n = 15), the majority (60.0%, n = 9) identified “monthly” as the optimal training frequency. “Biweekly” use was the second most common preference (26.7%, n = 4), followed by “once every few months” (13.3%, n = 2).

Discussion

Role of multimaterial simulation in otologic education

Multimaterial 3D-printed temporal bone phantoms offer a practical and effective alternative to cadaveric dissection for otologic training. By combining a calcium-enriched PLA matrix for osseous structures with contrasting polymers for neurovascular structures, the model allows clear visualization of critical “no-go” zones while approximating the mechanical resistance of cortical drilling. High adoption intent among residents (93.8% likely to use) underscores the demand for accessible, high-fidelity simulation tools. Although cadaveric specimens remain the gold standard due to unmatched anatomic fidelity and tissue realism, constraints such as cost, limited availability, regulatory considerations, and the inability to repeatedly practice delicate procedures underscore the need for complementary training modalities.^[7] At a material cost of $6.13 per unit and minimal postprocessing time, these phantoms provide a scalable, low-cost platform for developing spatial orientation, procedural sequencing, and psychomotor skills before cadaveric or intraoperative training. Integrating this model into a formal residency curriculum could follow a tiered, spaced-repetition approach. For example, novice trainees could complete monthly, modular phantom dissections focusing on foundational anatomy before advancing to scarce cadaveric resources.

Anatomic and visual fidelity

Overall anatomic accuracy and macroscopic appearance received favorable evaluations, with mean ratings of 3.56 and 3.62, respectively, between “moderately realistic” and “very realistic.” Junior residents (PGY-2 and PGY-3) consistently assigned higher ratings than senior residents (PGY-4 and PGY-5), reflecting the influence of prior experience on critical evaluation. Anatomic realism was identified as the highest-priority domain (importance 4.56), suggesting that accurate spatial representation is central to the educational value of the phantom. Visual fidelity, including color-coded neurovascular structures, was frequently highlighted in subjective feedback, enhancing the model’s utility as a learning tool by allowing trainees to identify surgical landmarks and practice safe dissection techniques.

Discrepancy between anatomic and haptic fidelity

A notable finding was the inverse relationship between resident experience and perceived realism. Senior residents, with more refined mental models of temporal bone mechanics, assigned lower ratings to tactile feedback (mean 2.67 to 3.00), a pattern consistent with the “expert critique” phenomenon described in simulation literature.^[33–35] Although the SimuBone material generated realistic dust and chips, a major improvement over the melting behavior of standard PLA, it received lower aggregate realism scores (3.06) compared with anatomic appearance (3.62). This suggests that while the current material formulation successfully mimics the visual mechanics of drilling (chip formation), it may not yet fully capture the tactile nuance of distinguishing variable bone densities (eg, the transition from mastoid cortex to trabecular bone).

For instance, PGY-4 residents assigned the lowest tactile ratings (mean 2.67), highlighting a specific limitation in the material’s ability to reproduce the “crunch” of cancellous bone. However, despite these limitations, most senior residents indicated a likelihood of phantom use, suggesting that precise haptic replication is not essential for training value. For more experienced trainees, the phantom’s utility appears to lie in high-level procedural rehearsal and anatomic navigation rather than pure haptic replication.

Cost-effectiveness and scalability

The low fabrication cost and compatibility with consumer-grade 3D printing hardware position multimaterial phantoms as a scalable solution for diverse training environments. Commercial synthetic temporal bones range from $200 to $300 per unit, while cadaveric specimens often exceed $500, emphasizing the financial advantage of this approach.^[36,37] This is particularly relevant for global health initiatives or smaller residency programs.^[38,39] Local manufacturing reduces dependence on centralized suppliers and may expand access for programs with budgetary or geographic constraints.

Limitations and future directions

Several limitations warrant discussion. The study cohort was relatively small (n = 16), which limited the statistical power for granular subgroup analyses across individual PGY levels (eg, the PGY-1 group). In addition, participants were recruited from regional programs, potentially introducing selection bias and limiting the generalizability of the findings to other training environments. Furthermore, the evaluation was also conducted within a cross-sectional, 2-day temporal bone course, which precluded prospective characterization of a longitudinal learning curve. Hence, the precise number of practice hours required for optimal skill transfer remains undefined, and the practice volume necessary for proficiency warrants investigation.

The evaluation also relied on subjective Likert-scale assessments, introducing potential observer bias, novelty effects, or a desire to please investigators. While anonymous survey collection and the more critical ratings provided by senior residents help mitigate these concerns, subjective data alone are insufficient to fully assess the phantom’s true utility. Triangulation with objective measures of skill transfer (eg, mastoidectomy completion time, volumetric accuracy of residual bone removal, blinded grading of anatomic landmark identification) would provide stronger evidence for educational impact.

In addition, although participants proceeded to cadaveric dissection after phantom-based training, this study relied on subjective survey ratings anchored to cadaveric experience and did not include direct, quantitative comparisons of performance. Standardized performance metrics and blinded assessment of surgical tasks, focusing on the avoidance of critical structures and procedural efficiency, will be implemented in forthcoming studies to allow rigorous, side-by-side evaluation and to determine whether phantom-based practice translates to improved cadaveric or operative performance. Assessment of haptic fidelity was similarly limited to subjective user feedback. Standardized, instrumented biomechanical testing was not performed to quantify drilling resistance, force profiles, or thermal degradation thresholds relative to native cortical and trabecular bone. Material optimization efforts should integrate quantitative biomechanical profiling using force-torque sensors, high-speed video analysis of chip formation, or instrumented surgical drills to objectively reduce the gap between synthetic and native bone characteristics.

Bench-top stabilization emerged as a practical challenge, with several participants noting that lateral supports were insufficient for vigorous drilling without supplemental clamping. Future iterations should incorporate standardized mounting interfaces compatible with common temporal bone holders to improve usability. Irrigation and debris management also remain areas for improvement, as particulate accumulation occasionally hindered visualization. Optimizing filament composition or infill patterns could address this issue. Finally, the educational implications of color-coded anatomy warrant further investigation. While subjective feedback suggests improved landmark recognition, formal studies are needed to determine whether reliance on color cues facilitates or impedes skill transfer to real surgical conditions.

Beyond training, this multimaterial framework has potential translational applications in preoperative planning and patient-specific surgical rehearsal. High-resolution CT-based segmentation can be adapted to produce patient-specific phantoms, enabling surgeons to rehearse complex or anatomically variant cases, anticipate procedural challenges, and refine surgical approaches before the operating room.^[40] Integration with image-guided navigation systems could allow for validation of registration workflows and the development of preoperative strategies. Moreover, the platform could serve as a testbed for evaluating novel surgical instruments, drill settings, or robotic-assisted approaches, providing a reproducible, low-risk environment for innovation.^[41,42] Future work should explore these applications, alongside longitudinal studies assessing whether preoperative rehearsal using multimaterial phantoms translates to improved operative efficiency, safety, and outcomes.

Conclusions

The multimaterial 3D-printed temporal bone phantom provides a cost-effective, anatomically accurate, and educationally valuable tool for otologic surgical training. While further material optimization is required to fully replicate the haptic complexity of human bone, the current model bridges the gap between virtual simulation and cadaveric dissection, offering a scalable platform for the acquisition of essential surgical skills.