A, The MeshLab open-source software (ISTI-CNR) rectangular selection tool was used to select large volumetric areas of mesh faces, which could then be removed. B, The Z-Painting tool in MeshLab was used to precisely select individual mesh faces for removal to isolate the desired structure (mandible). The arrow represents the mouse cursor.
The commercial model was from Materialise. Tooth roots and alveolar nerve canals are visible as brown colorations in the commercial models.
Measurements were used to assess for degree of variability between in-house, low-cost mandibular models and commercially produced models. Gn/id indicates the distance from gnathion to infradentale; Go-cdl, the distance from gonion to condyle; Go-cor, the distance from gonion to coronoid; Go-sig, the distance from gonion to sigmoid notch; Icdl, intercondylar distance; and Igo, intergonial distance.
The skull model demonstrates multiple-suture craniosynostosis after the first of 2 surgical procedures for a staged repair. Note the degree of detail and complexity attainable with low-cost, entry-level technology.
eFigure 1. Neonatal Mandible Models
eFigure 2. Decision Tree Analysis of the Choice Between Vendor Modeling Services and Low-Cost, Entry-Level, In-house Additive Manufacturing
Customize your JAMA Network experience by selecting one or more topics from the list below.
Legocki AT, Duffy-Peter A, Scott AR. Benefits and Limitations of Entry-Level 3-Dimensional Printing of Maxillofacial Skeletal Models. JAMA Otolaryngol Head Neck Surg. 2017;143(4):389–394. doi:10.1001/jamaoto.2016.3673
Can an in-house, entry-level 3-dimensional printed model serve as an alternative to high-end commercial products and be justified financially and clinically?
This case series found that entry-level, in-house production was comparable to high-end vendor modeling for surface anatomy and physical resilience, superior for cost and speed of production, variable for nerve canal visibility, and inferior for sterilizability, virtual surgical planning options, and tooth root visibility.
For purposes of uncomplicated surgical planning, education, and plate bending, in-house entry-level 3-dimensional printing is achievable with minimal training and at a significantly reduced cost compared with vendor products, which may easily justify its use in an otolaryngologic practice.
A protocol for creating exceptionally low-cost 3-dimensional (3-D) maxillofacial skeletal models does not require proficiency with computer software or intensive labor. Small and less affluent centers can produce models with little loss in accuracy and clinical utility.
To highlight the feasibility and methods of introducing in-house, entry-level additive manufacturing (3-D printing) technology to otolaryngologic craniofacial reconstruction and to describe its clinical applications and limitations, including a comparison with available vendor models.
Design, Setting, and Participants
This case series of 6 models (3 pairs) compared cost, side-by-side anatomical model fidelity, and clinical versatility using entry-level, in-house 3-D pediatric mandible model production vs high-end, third-party vendor modeling, including a review of the literature. Comparisons were made at an urban pediatric otolaryngology practice among patients who had previously undergone pediatric craniofacial reconstruction with use of a commercially produced medical model for surgical planning. Each vendor model had been produced using computed tomographic imaging data. With the use of this same data source, in-house models were printed in polylactic acid using a commercially available printer. Data were collected from November 1 to December 30, 2015.
Main Outcomes and Measures
Models created from these 2 methods of production were assessed for fidelity of surface anatomy, resilience to manipulation and plate bending, cost of production, speed of production, sterilizability, virtual surgical planning options, and alveolar nerve canal and tooth root visibility in mandibles.
For the quantitative comparisons between in-house models (1 neonatal, 1 pediatric, and 1 adult model) and their commercial counterparts, the mean value of 7 independent measurements was analyzed from each of 3 model pairs. Caliper measurements from models produced through entry-level, in-house manufacturing were comparable to those taken from commercially produced counterparts, suggesting an acceptable degree of accuracy (0.54 mm; 95% CI, 0.36-0.72 mm). Fixed costs for in-house production included acquiring an entry-level printer (retail $2899) and an annual software subscription ($699 per year). After purchase of these initial assets, the printing cost for an in-house mandible was approximately $90, with 98% of that cost related to labor. Physical qualities of entry-level, in-house models such as nerve canal visibility, tooth root visibility, and sterilizability were inferior compared with commercially-produced stereolithic renderings.
Conclusions and Relevance
This low-cost method of in-house, entry-level 3-D printing of straightforward, skeletal models may suit a general otolaryngology practice that performs maxillofacial reconstruction. Although commercial modeling offers several unique features, such as sterilizable materials and advanced virtual planning, in-house modeling also produces renderings with high fidelity, which may be used as tools for education and surgical planning, including preoperative plate bending.
Level of Evidence
Three-dimensional (3-D) printing is a rapidly advancing technology that is increasingly being adopted by medical professionals for clinical and research purposes. Three-dimensional printing, also termed additive manufacturing, refers to any number of techniques that construct solid objects in a layer-by-layer fashion, based on computer-aided design software models. Within otolaryngology, applications for additive manufacturing currently include patient counseling, surgical planning, design and creation of custom implants and prostheses, research, and tools for medical education and training, among others.1-3
The fabrication of models to scale is known as rapid prototyping and can be achieved using a wide variety of materials ranging from plastics and metals to synthetic biomaterials and even living cells.4,5 Printing methods most commonly include inkjet printing, fused deposition modeling, stereolithography (STL), and selective laser sintering.6 In clinical practice, rapid prototypes are typically obtained through a third-party vendor.7 These companies use state-of-the-art machines to construct high-resolution models, cutting guides, metal plates, and implantable prostheses; the cost of such technology can reach $500 000 or more.
As an alternative, a variety of low-cost, entry-level 3-D printers are available to the general consumer. These machines range in price from hundreds to several thousands of dollars and generally print with thermoplastic or bonded powder compounds. Recent research has explored the educational value of printing with such materials, including creating low-cost temporal bone models as alternatives to cadaveric specimens for surgical training.8-12 Another application for basic modeling includes the use of computed tomographic data to produce an accurate but nonsterile model of a patient’s maxillofacial skeleton. Using the model’s surface anatomy for accurate preoperative bending of reconstruction plates may optimize aesthetic outcomes after reconstructive surgery.13 Despite the low cost and apparent value of plastic or powder printing, these models are also often ordered from vendors instead of being produced in-house.9,13
In this report, we describe our in-house production process of 3-D printed pediatric mandible models. We compare commercially manufactured medical models with our markedly lower-cost thermoplastic models, which were constructed with entry-level additive manufacturing technology that was already in use within the information technology department of our medical center. We aim to highlight the opportunity for smaller centers and hospitals with limited resources to create accurate skeletal models, which are suitable for most clinical purposes.
Patients who had previously undergone pediatric craniofacial reconstruction were identified for review, and those with a commercially produced medical model were selected for comparison purposes. The institutional review board of Tufts Medical Center, Boston, Massachusetts, granted a Health Insurance Portability and Accountability Act waiver of research authorization and a waiver of consent for this study.
Data were analyzed from November 1 to December 30, 2015. Preoperative computed tomographic data with a section width of 1 mm had been supplied to 1 of 2 commercial vendors (Materialise or 3D Systems) to produce a high-fidelity model of the skull, mandible, or maxilla. The data from these scans were accessed in a standard digital imaging and communications in medicine (DICOM) format. To produce our own model, each DICOM file was imported to imaging software (OsiriX; Pixmeo SARL) and rendered using the 3-D surface tool with the first surface variable set to the predefined computed tomographic–bone pixel value of 500. This process removed soft-tissue elements, and the resulting surface mesh was exported in STL format. Varying selection tools were used in MeshLab open-source software (ISTI-CNR) to select and precisely remove unwanted surface mesh near the desired product (Figure 1). In this study, we chose to use mandibles for comparison purposes. The surface mesh, now in the form of a mandible, was saved in STL format and printed with a fifth-generation 3-D device (MakerBot Replicator Desktop 3-D Printer; MakerBot Industries LLC) oriented with its inferior border resting on the build platform. This printer used fused deposition modeling technology to print in polylactic acid thermoplastic with a layer thickness of 0.1 mm. Suspended components are built on a supporting raft that is eventually broken off by hand and sanded after production. The in-house models were printed with natural color polylactic acid filament (MakerBot Industries LLC) using 10% infill to achieve a balance between translucency and physical strength. Side-by-side views of vendor models and our entry-level counterparts are shown in Figure 2 and eFigure 1 in the Supplement.
We compared internally produced model mandibles with their commercial counterparts to assess for several properties. To determine objective and quantifiable data comparing the 2 types of models, a series of standard measurements were taken from 3 commonly produced model types. Using industry-produced models as the criterion standard, we assessed 3 in-house prototypes (a neonatal mandible, a pediatric mandible, and an adult mandible). Using a dial caliper gauge (Peacock; Ozaki Manufacturing Co), we obtained 3 separate measurements of 7 predetermined distances on each anatomical model (Figure 3). We determined the mean of the 3 measurements and created 7 samples for each model to be used for analysis. To compare industry models with the in-house prototypes, we calculated the absolute difference of the mean for each of the 7 samples associated with each model. Data were entered in an Excel spreadsheet (Microsoft Corporation) and 95% CIs were determined; significance was set at P < .05.
Subjective measures, including model sterilizability, capacity for virtual surgical planning, visibility of tooth roots and alveolar nerve canals, perceived fidelity of surface anatomy, and resilience to manipulation during preoperative bending of the reconstruction plates, were considered. Finally, the speed of fabrication and estimated cost of production were examined for each model.
For the quantitative comparisons between in-house models and their commercial counterparts, the mean value of 7 independent measurements was analyzed from each of 6 models (3 pairs). To determine whether accuracy varied by the size of the model, we included a prototype of a neonatal mandible, a pediatric mandible, and an adult-sized mandible. The mean absolute difference in the 7 anatomical measurements was 0.39 mm (95% CI, 0.16-0.62 mm) for the neonatal model, 0.68 mm (95% CI, 0.30-1.05 mm) for the pediatric model, 0.55 mm (95% CI, 0.12-0.99 mm) for the adult model, and 0.54 mm (95% CI, 0.36-0.72 mm) among all 3 models.
Entry-level, low-cost, in-house production was inferior to commercial modeling for sterilization capability, capacity for virtual surgical planning, visibility of the tooth roots, and availability of printed hardware and cutting guides. Alveolar nerve canal visibility was variable among the low-cost in-house models; however, the scale, surface topography, anatomical detail, and physical resilience appeared to be comparable to those of the commercial products. The cost and speed of in-house production were superior to those of commercially ordered mandible models.
The cost of a commercially produced mandible varied on a case-by-case basis but was estimated by vendors to exceed $1000. The charges for virtual surgical planning services also started at approximately $1000. At present, fixed costs for in-house production include acquiring the 3-D printer (retail, $2899) and an annual software subscription (OsiriX) for medical use ($699 for 1 year). Excluding the cost of labor, the total variable cost of printing for a mandible ranged from $0.94 to $1.83, depending on the amount of material needed for printing, which varied by the size of the mandible and the density of infill. Cost for polylactic acid plastic filament was $50 per kilogram, with 17 to 33 g required per model.
The turnaround time for in-house production was 1 day, requiring approximately 20 to 60 minutes for software processing and printer preparation. Mean production time ranged from 6 to 12 hours for printing at 10% infill. Postproduction modification (any sanding or removal of support elements) and delivery of the model was approximated at 60 minutes. Because printing was typically performed overnight or during the day without supervision, print time was not considered a billable part of labor costs. Assuming the production of 1 model per week, the approximate labor cost amounted to 2 hours per week and may be estimated as one-twentieth of the workload of a single, fulltime information technology staff member. For the purposes of this study, we based that cost on the extrapolated hourly rate of a senior information technology employee at our hospital ($45/h × 2 hours = $90.00). Total costs per mandible model ranged from $90.85 to $91.65. A decision tree analysis of the choice between vendor modeling services and in-house additive manufacturing is given in eFigure 2 in the Supplement.
Vendor model turnaround time was generally 3 business days if 1-day shipping was used. If a virtual planning session and customized sterile cutting guides were added to the order, an additional 2 to 3 business days were required.
We herein describe the use of readily accessible, entry-level additive manufacturing technology to construct rapid prototype maxillofacial models for clinical use. Caliper measurements comparing these low-cost models with commercial counterparts suggest an acceptable degree of accuracy (0.54 mm; 95% CI, 0.36-0.72 mm). Although a rapidly growing body of literature demonstrates the expanding capabilities of in-house additive manufacturing technology, most of these reports are from academic institutions and regional referral centers where a combination of clinical demand and academic funding have allowed for significant capital expenditure and investment in personnel dedicated to the building of a 3-D printing program. This report seeks to demonstrate that in-house additive manufacturing for medical purposes using entry-level technology can serve as a lower-cost alternative without compromising the anatomical accuracy needed for simple reconstruction and planning. High-quality models may be produced even with the most basic technology (Figure 4). In sharing our experience, we hope to inspire smaller institutions and centers located in settings with lower levels of resources to enter the sphere of in-house additive manufacturing technology.
At present, the most significant advantage of commercial additive manufacturing through a third-party vendor lies in the numerous surgical planning options available and the vendor’s ability to incorporate these plans into a 3-D rendering. Virtual planning requires proficiency in software and is beyond the ability of a basic information technology department staff member. Commercial manufacturers often have employees dedicated to these tasks. Commercial products for operative use may include the additional option of operative cutting guides and hardware that can be sterilized owing to the use of high-end processes that print in autoclave-resistant polymers and metals.6 Entry-level thermoplastic printers do not have this advantage.
Although vendor models are unique in their capability for sterilization, the value of an intraoperative reference that may be brought directly into the surgical field (rather than simply examined adjacent to the field with gloves) may be of limited benefit. For purposes of patient counseling, medical education, preoperative bending of reconstruction plates, and intraoperative reference, in-house models have value and are comparable in quality and accuracy to commercial models.14 Studies cite as much as a 1- to 3-hour reduction in operative time when reconstruction plates and templates have been molded before the procedure.15,16 Time saved in the operating room reduces cost and patient exposure to general anesthesia.
An additional distinction of commercial models is that internal structures such as tooth roots and nerve canals are colored and, given the STL material’s semitransparency, clearly visible. We note that the visibility of alveolar nerve canals was variable among in-house plastic models and seemed to depend on canal diameter. With a layer thickness of 0.1 mm, our models should show even the narrowest channels if they are defined in the original 3-D surface mesh. We visualized patent alveolar nerve canals on all the larger rapid prototypes.
Although we did not perform quantitative mechanical stress testing, very little torque was applied to the models during preoperative bending of the plates, and we believe that they are resilient to firm handling. In addition, we had the option to print with a higher infill to achieve greater density at the expense of limiting translucency.
Expertise with computer-aided design software seems to be a significant variable in the disparity between high- and low-end products, virtually and physically. With training, we could manually segment the surface mesh to ensure a patent nerve canal is sent to the printer. Studies using similar rapid prototyping technology in the production of temporal bone models describe such manual segmentation to properly print small bony components such as ossicles.10,17 This skill would be relatively simple to learn.
The gap in virtual surgical planning may, in theory, be narrowed by in-house proficiency with computer-aided design software. Furthermore, slightly more expensive printers than ours can print in multiple colors (based on software input) and may be instructed to print key structures, such as alveolar nerve canals, using a visible pigment. However, unlike simple segmentation, the considerable complexity and clinical implications of such a design would require far greater surgeon input and oversight. In addition, a substantial time commitment by a software technician would be necessary for initial training and case-specific design. Our in-house production protocol resulted in a next-day mandible model, which is faster than commercial avenues by at least 1 day. This expedited production keeps in mind that our variables were set at a 10% infill. Increasing infill would increase print time and decrease translucency but provide greater model density and strength. Our models felt firm in hand and were resilient to manipulation without special precautions, although we did not conduct quantitative strength testing.
This study is limited by a paucity of objective data, limited to measuring the fidelity of model dimensions as they compare to one another. We did not perform formal strength testing, and all models were created by the same technologist and printed on the same printer. A variety of entry-level printers are available, and we only used the model in this study. We therefore cannot determine whether variation exists among models fabricated by different technologists or printed using different devices.
Assuming that a commercially produced mandible model costs $1000 and for most purposes is clinically comparable to a model produced internally, in-house production using entry-level 3-D printing technology and software pays for itself after 4 prints. Although commercial modeling offers several unique features, such as sterilizability and advanced virtual planning tools, entry-level printing also provides acceptably accurate models for use in simple reconstruction planning and plate bending. Other important benefits may include patient counseling and medical education. The number and complexity of cases seen at a given hospital may justify institutional investment in a 3-D printer, especially if commercial models are routinely ordered.
Corresponding Author: Andrew R. Scott, MD, Division of Pediatric Facial Plastic and Reconstructive Surgery, Department of Otolaryngology–Head and Neck Surgery, Floating Hospital for Children–Tufts Medical Center, 800 Washington St, PO Box 850, Boston, MA 02111 (firstname.lastname@example.org).
Accepted for Publication: October 1, 2016.
Published Online: January 5, 2017. doi:10.1001/jamaoto.2016.3673
Author Contributions: Drs Legocki and Scott had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Legocki, Scott.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Legocki.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Legocki, Scott.
Administrative, technical, or material support: All authors.
Study supervision: Scott.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.
Meeting Presentation: This paper was presented at the Triological Society Combined Sections Meeting; January 23, 2016; Miami Beach, Florida.