Radiological Society of North America (RSNA) 3D printing Special Interest Group (SIG): guidelines for medical 3D printing and appropriateness for clinical scenarios
3D Printing in Medicine volume 4, Article number: 11 (2018)
Medical three-dimensional (3D) printing has expanded dramatically over the past three decades with growth in both facility adoption and the variety of medical applications. Consideration for each step required to create accurate 3D printed models from medical imaging data impacts patient care and management. In this paper, a writing group representing the Radiological Society of North America Special Interest Group on 3D Printing (SIG) provides recommendations that have been vetted and voted on by the SIG active membership. This body of work includes appropriate clinical use of anatomic models 3D printed for diagnostic use in the care of patients with specific medical conditions. The recommendations provide guidance for approaches and tools in medical 3D printing, from image acquisition, segmentation of the desired anatomy intended for 3D printing, creation of a 3D-printable model, and post-processing of 3D printed anatomic models for patient care.
In 2016, the Radiological Society of North America (RSNA) approved a proposal to create the Special Interest Group on 3D Printing (SIG). This document fulfills two of the original SIG goals: to provide recommendations towards consistent and safe production of 3D printed models derived from medical images, and to describe a set of clinical scenarios for 3D printing is appropriate for the intended use of caring for patients with those medical conditions. This project also fills a previously unmet need for practice parameters/guidelines regarding the clinical service of anatomic modeling (3D Printing) described for proposed new billing codes, including those for the American Medical Association. These practice parameters and recommendations are not intended as comprehensive standards but do reflect several salient aspects of clinical anatomic modeling and appropriateness. The guidelines subcommittee of the SIG will maintain and devote the time and effort necessary to continually develop and update these recommendations. This subcommittee is comprised of volunteer members of the SIG who form the writing group of this document.
Consensus methodology recommendations
Appropriateness consensus guideline generation
The Special Interest Group has initiated the quality and safety scholarship to identify those clinical situations for which 3D Printing is considered an appropriate, and not appropriate, representation of the data contained in a medical imaging examination. These documents highlight appropriateness of medical 3D printing for clinical utilization, research, scientific, and informational purposes. This work is loosely modeled after the American College of Radiology Appropriateness Criteria® [553, 554] in that the guidelines committee uses an evidence-based approach at scoring. Consensus among members is used when there is a paucity of evidence.
Each category was led by a separate writing group, composed of a small group of experts in that domain of medical imaging and 3D printing. The SIG Executive committee, led by the Guidelines Chairperson, formed the review panel. Ratings were generated via by a vote of Special Interest Group members at in-person meetings. The results of the ratings follow the following 1–9 format (with 9 being the most appropriate):
1–3, red, rarely appropriate: There is a lack of a clear benefit or experience that shows an advantage over usual practice.
4–6, yellow, maybe appropriate: There may be times when there is an advantage, but the data is lacking, or the benefits have not been fully defined.
7–9, green, usually appropriate: Data and experience shows an advantage to 3D printing as a method to represent and/or extend the value of data contained in the medical imaging examination.
The supporting evidence was obtained through structured PubMed searches, as detailed in the Appendix. In rare circumstances, supporting literature was recommended directly by the members of the committee and was explicitly identified outside of the structured PubMed search results.
A subset of applications of 3D printing, including in congenital heart, vascular, craniomaxillofacial, musculoskeletal, genitounirary, and breast pathologies was selected for detailed review. All final components of this section were vetted and approved by vote of Special Interest Group members at several face-to-face meetings including on August 31 (Silver Spring, MD) and December 1, 2017 (Chicago, IL) as well as via internal posting on the SIG member intranet.
Consensus methodology recommendations
Medical image acquisition
The most common medical imaging modalities for 3D printing are computed tomography (CT) and magnetic resonance imaging (MRI); however, any 3D imaging dataset including sonography (e.g., echocardiography) may be utilized as input data for segmentation. The international standard format for these imaging files is Digital Imaging and Communications in Medicine (DICOM). At this time, DICOM images are not routinely sent directly to a 3D printer for printing, so medical images are segmented and converted to a file type that is recognized by 3D printers. Common file types include Standard Tessellation Language (STL), OBJ, VRML/WRL, AMF, 3MF, and X3D. Once this functionality is implemented by 3D printing vendors, picture archiving and communication system (PACS) vendors, and at the point of care facility, it will allow 3D files in the form of STLs, for example, to be stored in a patient’s medical record.
Spatial resolution and slice thickness
Medical imaging data should have sufficient spatial resolution to accurately represent the anatomy to be modeled. The spatial resolution of an imaging method refers to the smallest resolvable distance between two different objects or two different features of the same object. Low spatial resolution techniques will be unable to differentiate two adjacent structures that are close together and have similar tissue properties. When the intent to produce a 3D model is known prior to a medical imaging procedure, the image acquisition should be tailored so that the anatomy in the intended 3D model can be adequately visualized. The optimal spatial resolution will depend on the anatomy being imaged.
Slice thickness, which influences the spatial resolution and image noise (discussed in the next section), can also be optimized depending on the intended use. In general, this means that the smallest anatomy of interest should be captured on multiple sequential DICOM images of a particular series. For example, if the anatomy of interest measures 3 mm, it would be desirable for this anatomy to be captured on at least 3 sequential image slices; therefore, the slice thickness should be no greater than 1 mm, and preferably smaller. If images are acquired with a large slice thickness, stair-step boundaries may be seen in the 3D model.
For CT, in combination with scan distance, consideration may be given regarding collimation (the thickness of the X-ray beam) and overlap. Typically, the scan distance and collimation are the same; however, if the slice distance is smaller than the collimation, there will be an overlap which may lead to improved results. Cone Beam CT has technical differences with conventional CT, and often results in a lower patient radiation exposure and subsequently less image contrast that typical clinical CT images. Image artifacts and consistency of SNR throughout the scan can also limit studies. For MRI, voxels may be isotropic or rectangular solids and the resolution may be different in the three dimensions. The size of the voxel depends on the matrix size, the field of view (FOV), and the slice thickness.
In some clinical scenarios, there are patients for which suboptimal imaging data is available, but a separate acquisition is contraindicated. If superior spatial resolution is preferred and CT data is required, that benefit should be weighed against the risk of delivering more radiation to the patient.
Signal to Noise Ratio (SNR) and Contrast to Noise Ratio (CNR)
The SNR is a metric of image quality. A higher SNR, all else being equal, implies more trustworthy data for 3D printing. The CNR is the relationship of the signal intensity differences (the contrast) between two regions, scaled to noise. High contrast between various organs in the body is an important feature of medical imaging and is necessary to delineate structures for 3D printing. The SNR and CNR of images used for 3D printing should be comparable to, or superior to, those for “3D visualization”, defined as the comprehensive ensemble of manipulation of a volumetric data set for viewing on a 2D surface such as a computer monitor .
If the SNR and/or CNR are inconsistent, or suboptimal, the risks of inaccurate segmentation must be weighed against those of rescanning the patient. Regarding high noise data, a judgment call must be made to determine whether the segmentation operator is capable of delineating the data (e.g. in the case of a cone beam CT image series).
In CT, the X-ray tube voltage may also be adjusted to maximize the signal. A lower kV can be used to increase the enhancement of iodine contrast when building vascular models. In addition, the raw data reconstruction parameters selected may affect the appearance of specific anatomical structures. For example, the reconstruction kernel (image filter) impacts both the spatial resolution and image noise, which must be balanced, based on the application. Typically, kernel options range from “sharp” to “smooth.” Sharpening filters increase edge sharpness at a cost of increasing noise while smoothing filters reduce noise content in images by also decrease edge sharpness. For models with fine structures, such as the temporal bone, a sharp kernel is preferred; and for larger, low contrast models, a smooth kernel is more appropriate. CT is considered the imaging modality of choice for bone imaging and is often used to produce 3D anatomical models of hard tissue structures such as bone. In MRI, the SNR may be improved by performing a volume acquisition (at the expense of time), decreasing noise by reducing the bandwidth, altering the echo time or repeat time, increasing the FOV, decreasing the matrix size, or increasing the slice thickness.
The sub-volume of the imaging dataset that will be 3D printed is defined in this document as the printing Region of Interest (ROI). All medical images contain artifact, and image processing steps should be taken to minimize artifact. The ROI should be small enough to enable confident segmentation for 3D printing. There are cases for which medical interpretation is possible (see Image interpretation Section), but 3D printing can be limited by the presence of artifact, motion, or other spatial or noise limitations in DICOM images. When this is the case, we recommend that the model be annotated with documentation of those parts of the ROI where segmentation quality may be limited.
Medical images acquired for a clinical indication should be interpreted with the interpretation being incorporated into the patient medical record. The interpretation should include the ROI being considered for printing. Often, interpretation of the ROI incorporates 3D visualization to enable or enhance diagnosis. Examples of 3D visualization include multi-planar reformatting, maximum intensity projections, and volume rendering. Such interpretations are currently billable in the United States under CPT codes 76376 and 76377.
Image data preparation and manipulation
Image segmentation is necessary to create 3D printable files from medical images. The segmentation process, which subdivides medical images into anatomical regions, typically begins by importing a set of DICOM images into dedicated image post-processing software. Anatomical regions are selected using a combination of automated and semi-automated tools. Once the desired ROI for 3D printing has been selected, data is interpolated and a surface-based 3D model which describes the 3D geometry of that volume is calculated. To date, the most common, widely used, and accepted file format for medical 3D printed objects is the STL file.
STL files are composed of triangular faces, and the number of these faces can affect anatomical accuracy of a model. Each lab should determine the appropriate number of faces/triangles for their medical models to adequately represent anatomy. Operators should be aware of any reduction, smoothing, or further file manipulation or optimization within the segmentation software when creating and exporting the STL file.
The contours of the STL file should be routinely checked against the source medical imaging data; typical segmentation software packages allow the final STL to be re-imported and its contours displayed over the original DICOM images. This option can be used to verify the surface accuracy of an anatomical model STL file. Additional file formats noted above should also meet the same criteria.
Segmentation and Computer Aided Design (CAD) software
Medical image processing software is required to generate a file format amenable to 3D printing. The RSNA 3D printing SIG concurs with the FDA that software that has been favorably evaluated by the FDA be used to translate medical images into formats amenable to 3D printing for all aspects of patient care, defined by the SIG as all interactions with healthcare professionals, or patients and their families, related to medical care. The SIG recommends that software used for segmentation is FDA cleared to produce 3D Printed models suitable for diagnostic use, specifically using the FDA definition of diagnostic use and noting that FDA cleared software for 3D printed models will also include machines and materials validated for this intended use. At the time of manuscript submission, the FDA has approved one complete system, consisting of software through the printing process, for medical model production.
File storage and descriptors
Files stored within a repository should contain or be linked to a set of corresponding descriptors, including those pertaining to image acquisition and further imaging processing. Descriptors should be supported by standardized terminology from a consensus vocabulary; the SIG acknowledges that this vocabulary represents a current, unmet need. If the descriptors are not within digital files, this information should be otherwise archived.
Reference to file manipulation and alteration
Data from medical images undergo alterations in the design of the physical model. These changes have been categorized into Minor and Major alterations , with the latter generally representing changes that could impact clinical care. When modifications include major changes, the operator should verify that both the digital file and 3D printed model is labeled/identified appropriately.
Generation of 3D printed model
There are many different 3D printing technologies, each differing in the way that the final 3D printed model is created. When 3D printed models are generated from medical images, the resolution of the 3D printer should be equal to, or superior to that of the clinical images used to segment the model. Similar to the DICOM acquisition stage, it is preferable that printed layers be a multiplier of the smallest geometry of interest. For example, if the smallest anatomical object of interest on the 3D model is 1 mm, this object should be printed on at least 3 layers of the model. Due to the nature of medical models, and the need for sub-millimeter accuracy, a layer thickness of no more than one-third of a millimeter is recommended, and preferably less than or equal to one-eighth of a millimeter. In addition to the layer thickness of the 3D printing hardware, the in-plane (x-y) resolution should be known, with a target of less than one-quarter of a millimeter. The values above are global recommendations may not be applicable in all cases. If a model requires a higher or lower accuracy, these parameters should be modified accordingly.
The medical model should include a patient identifier or an internal unique identification number that can be tracked back to the patient and date of the image acquisition. Labels can be incorporated (3D printed) into the model itself. Labels should be externally attached to the model if size or location does not allow for printed labeling. Printed models are assumed to be of anatomic size (1:1) unless a scaling factor is otherwise noted. Additional identifiers such as model sidedness (left, right) should be noted, as appropriate. Institutional guidelines should be used to verify models are free of protected health information, or models are handled appropriately in accordance to Health Insurance Portability and Accountability Act (HIPAA) guidelines.
Post-processing printed models
Post-processing steps should not alter the intended morphology and desired accuracy of the part, but instead should only enhance the utility (including clarity and transparency) and/or durability of the model. It should be noted that finishing may slightly alter the dimensional accuracy of a part, but this variation should be minimal (or within the desired global accuracy of the part) and the benefits (for example: strength and clarity) should outweigh the dimensional change. All support materials and residual manufacturing materials and/or substances should be removed as completely as possible. If all supporting material is not capable of being removed, this should be noted and presented to the requesting provider. Should the model be damaged either during or after post-processing and cleaning, repairs should be performed in a manner that reconstitutes the quality to which the original model adhered. If these repairs are not possible, the model should be reprinted. Any damage should be noted to the provider and the option to reprint should be presented. Cleaning solution concentration and saturation levels should be monitored and maintained in accordance to manufacture recommendations.
The model should be inspected by the 3D printing laboratory before clinical use. For cases where the model may be limited by a known image artifact, the model will be noted with any areas of concern. Qualitative and/or quantitative measures to confirm that the 3D printed model matches the desired input data will be taken, including but not limited to expert subjective assessment and objective fitting to the original volume submitted for printing. This can be done on a per part basis, per build basis, or in accordance with an additional internal protocol of the 3D lab. Some examples of qualitative assessments could include comparing the model to a digital representation or printed picture of the model and inspecting the model for printing imperfections or inaccuracies. Some examples of quantitative inspections could include measurements of a test specimen, measurements of the model, or scanning and comparing the model back to the original DICOM data sets.
The U.S. Food and Drug Administration (FDA)
The U.S. Food and Drug Administration (FDA) ensures the safety and efficacy of personalized devices in the United States of America. 3D Printing falls under the auspices of The Center for Devices and Radiological Health (CDRH). There have been four FDA benchmarks related to 3D printing and medical devices from 2014 to 2018.
First, in October 2014, the FDA held a public workshop entitled “Additive Manufacturing of Medical Devices: An Interactive Discussion on the Technical Considerations of 3D Printing”. Second, the FDA published “Additively Manufactured Medical Products – The FDA Perspective” . Third, in December 2017, the FDA published “Technical Considerations for Additive Manufactured Devices” . This perspective included insights regarding 3D printing data manipulation and hardware for modeling patient-specific anatomy. Fourth, the FDA commented on the publication “Maintaining Safety and Efficacy for 3D Printing in Medicine” . This paper uses a similar, logical 3-step format of these consensus recommendations, and then develops different suggestions for regulatory models that depend on how much, if at all, the anatomical data is modified before 3D printing. On August 31st, 2017, the RSNA SIG and the FDA engaged in a joint meeting to discuss 3D printed anatomic models. The intended output of this meeting is a co-published white paper that will form the next benchmark.
Quality control program
Due to environmental factors and material properties, model morphology is expected to change over time. As part of a complete quality control program, 3D printers should undergo regular accuracy testing, including test prints, preventative maintenance, and recalibration [581, 582]. Laboratories may develop a process using a phantom to ensure regular quality standards for their printers. If the reference standard is known or assumed, mathematical operations  can be applied equally to those volumes in the ROI to determine the overall accuracy of the model, including not only potential manual errors from segmentation, but also generation of the final data set including digital post-processing steps such as smoothing.
Delivery and discussion with referring physicians
3D printed models represent an advanced form of communication of the data in medical images, and may include the summation of data from multiple sources. Extensive multidisciplinary teaching opportunities for 3D printing have been realized [584,585,586]. Physicians should have an opportunity to discuss the salient features and intended use of all models. Any concerns about the model or segmentation process, if not discussed previously, should be noted to the provider at the time of delivery. Where possible, annotations detailing critical points of model anatomy should be stored both within the digital record of the model, and physically placed on the 3D printed model. One example is annotation of a subtle fracture that may not otherwise be represented in either or both, the segmented, or the 3D printed model.
“Pre-operative planning” with 3D printing refers to virtual surgical planning (also called digital templating, digital surgical planning, virtual planning, computerized planning, computer-assisted surgical simulation). This detailed planning of the intervention occurs in the digital space. There are times when the simulation itself is the end product, and the interventionist acquires valuable information regarding patient anatomy and medical devices to be used to increase confidence and knowledge before surgery. For these cases the digital plan is transferred to patient care by way of 3D printed templates, guides, or models. This type of planning usually involves major changes to the digital model while utilizing original patient contours. This necessitates the systematic application of the 3D printing recommendations outlined above to the models used for virtual surgical planning as a minimum requirement.
Material biocompatibility, cleaning, and sterilization
For anatomical models and surgical guides/templates/jigs potentially entering a surgical field, material biocompatibility, cleaning, and sterilization are vitally important. The details are beyond the scope of this document. However, biocompatibility of materials depends on several factors including base material, the 3D printing process (and any variations), any post-processing techniques, and hospital cleaning and sterilization methods and requirements. Manufacturers should provide cleaning recommendations and specifications for materials which have been formally tested for biocompatibility and sterility, and these specifications should be followed by the facility. Additional internal sterilization policies may exist depending on the hospital.
Appropriateness of 3D printing (anatomic modeling) for selected clinical scenarios
This section provides evidence-based guidelines, supplemented by expert opinion when there is a paucity of peer-review data, to define and support the use of 3D printing for patients with a variety of conditions, including congenital heart, vascular, craniomaxillofacial, musculoskeletal, genitounirary, and breast pathologies (Table 1).
Reviews that include the types of 3D printers commonly used in medicine have been published [563, 584]. Regarding image post-processing and software, several tutorials are available for step-by-step training. The following discussion includes the specific descriptions from the SIG writing group for each clinical group of clinical scenarios considered for appropriateness.
Congenital heart disease
Congenital heart diseases (CHD) are the most common significant birth defects. Substantial literature supports the benefit of 3D printing for patients with congenital heart disease [1,2,3,4,5,6,7]. Regarding improved outcomes, precise preoperative understanding of the complex anatomy from a printed model may obviate or shorten lengthy exploration, and therefore operation and cardiopulmonary bypass time can be reduced.
These recommendations utilize and conform to the CHD nomenclature defined by the European Association for Cardio-Thoracic Surgery / Society of Thoracic Surgery (EACTS-STS) version of the International Pediatric and Congenital Cardiac Code (IPCCC), except as where noted otherwise. The clinical scenarios defined by the IPCCC include the following: Septal Defects, Pulmonary Venous Anomalies, Cor Triatriatum, Pulmonary Venous Stenosis, Right Heart Lesions, Left Heart Lesions, Single Ventricle, Transposition of the Great Arteries, DORV, DOLV.
Structured searches were performed using the US National Library of Medicine (PubMed), which enabled the querying and retrieval of appropriate clinical documents supporting the appropriateness of 3D printing-enabled technologies for each specific diagnosis. The search results were reviewed by experts and some references were removed and some were relocated to different categories. As noted above, references outside of the structured searches were added but noted and approved by the writing group. As a general rule, the benefits of 3D printing to define and rehearse an intervention increase with the overall degree of complexity of disease.
The International Classification of Diseases, Tenth Revision (ICD-10)  descriptions and categorization were used to categorize the clinical scenarios for rating craniomaxillofacial conditions. Four major groups were used as the starting point; 1) Craniomaxillofacial Trauma, 2) Congenital Malformations, 3) Acquired/Developmental Deformities and 4) Neoplasms. Further sub-groups were formed underneath the major groupings. Additional clarification for “simple” versus “complex” diagnoses within a particular group was given based on inherent differences in appropriateness ratings between subgroups of patients in these groups. Further language describing each diagnostic grouping helps describe the difference between a simple and a complex case in each subcategory.
Structured searches were performed using the US National Library of Medicine (PubMed), which enabled the querying and retrieval of appropriate clinical documents supporting the appropriateness of 3D printing-enabled technologies for a specific condition. The search results were reviewed by experts and some references were removed because they were not relevant. A small number of references were added because they were found to be relevant, but not appearing using the stated search string. As noted above, these were vetted by the writing group before inclusion. Clinical scenarios that were only dental or only brain have not been included. The authors recognize that these include many important clinical scenarios of for 3D printing, and the goal is to include them in upcoming documents.
Craniomaxillofacial (CMF) conditions for the purposes of this document encompass several different surgical specialties all working in the head and neck area with both pediatric and adult patients. These include oral and maxillofacial surgery, craniofacial surgery, plastic surgery, microvascular surgery, pediatric neurosurgery and otolaryngology. Use of 3D printing-enabled technologies to aid clinical care in the craniomaxillofacial area has been seen from the very advent of 3D printing in the late 1980s [556, 557]. Even before the commercialization of stereolithography there were surgeons, engineers and researchers figuring out more manual ways of converting medical imaging datasets into 3D models . The fit seems clear, CMF surgery has both a functional component, and for most cases an aesthetic component, where the form carries importance along with the functional restoration. In the CMF arena, the use of anatomical models of anatomy is primarily derived from CT and MRI datasets, and also from an increasing volume of cone beam CT datasets. Patient-specific anatomical models are the baseline, but for many of these applications the value of these technologies has been found in either, a) patient-matched implants (for instance temporomandibular joint reconstruction), or b) virtual surgery combined with templates and guides (for instance orthognathic surgery). The scenarios to follow were thought of in this way, with some of them relying heavily on anatomical models alone and some of them relying with increasing importance on the role that digital planning combined with patient-matched implants or templating is playing.
The genitourinary conditions have been organized anatomically, recognizing that common genitourinary interventions are largely based on anatomic considerations. The complication rate after major genitourinary surgeries is reflected in the complexity of the lesion. For example, more complex kidney tumors are associated with longer operative times, warm ischemia times, and greater blood loss . High kidney tumor complexity can also be correlated to the risk of major postoperative complications requiring a secondary intervention .
There is a growing body of literature that supports the benefits to patients from 3D printed models. Specifically, 3D printed models may improve comprehension of anatomy and facilitate pre-surgical planning for complex surgical cases, ultimately reducing operation times and improving patient outcomes.
This document describes and provides rating for the clinical scenarios related to 3D printing of genitourinary pathology [561, 562]. Structured searches were performed using the US National Library of Medicine (PubMed), which enabled the querying and retrieval of appropriated clinical documents supporting the appropriateness of 3D printing for a specific diagnosis. As a general rule, the benefits of 3D printing to define and rehearse a genitourinary intervention increases with the overall degree of complexity of the pathology that is represented by the physical model based on a medical imaging study performed in a radiology department.
The role of 3D printed models in addressing musculoskeletal pathologies can vary depending on a specific clinical scenario, ranging from aiding in informed consent to use in preoperative planning. Custom fixation plates, surgical osteotomy guides and implants can also be generated from 3D data, allowing for virtual surgery and design of a custom implant that is modeled after the contralateral healthy side. In addition, mock surgeries can be performed on the physical 3D models, allowing for more intuitive problem solving and measurements preoperatively. Such planning alters surgical management for some patients, either by delaying intervention, or by suggesting an alternative approach. Pre-surgical planning can also decrease operating room time and the number of devices and tools that need to be tried and subsequently wasted and/or re-sterilized. In this sense, 3D printing has proven useful for demonstrating musculoskeletal pathology and for planning interventions.
Based on the accumulating evidence, the use of 3D printed models can positively impact numerous metrics associated with musculoskeletal interventions, including patient and physician satisfaction, operative time, blood loss, and the various direct and indirect costs associated with patient-centered decision making regarding management of complex disease. At present, the musculoskeletal pathologies with potential and established 3D printing-enabled management have been broadly categorized into fractures, chronic osseous abnormalities, degenerative disorders, neoplastic pathologies, scoliosis, and miscellaneous specific applications including ligamentous injury and heterotopic ossification.
3D printing has been shown to be useful for understanding the vascular anatomy, evaluation of hemodynamics, treatment planning (surgical and endovascular) as well as preclinical testing of devices. It has also been used for medical education and procedural training on vascular models [563,564,565,566]. There are several clinical scenarios for which 3D printing has been used in the care of patients with vascular disease. Because of the nature of vascular pathology, dissection, aneurysm, and stenosis are often treated with medical management and “watchful waiting”; most patients follow this algorithm, and there is little to no role for 3D printing. However, some patients have a clinical presentation and non-invasive tests that warrant intervention, while others progress from watchful waiting to planned intervention. For many of these patients, 3D printing is appropriate. Of note, coronary 3D printing, and cardiac printing in general falls outside the scope of this document. These clinical scenarios will be discussed in future documents.
Most aortic dissections are treated medically, and for these patients there is no indication for 3D printing. However, 3D printing may be appropriate for planning intervention in complex dissections, and in particular dissections that also have enlargement. Models have been used for planning and simulation of stent deployment . Simulation on models can help in identifying the best projections for angiography, best catheter and wire combinations to navigate the anatomy, in for determining appropriate balloon and stent size as well as position.
Endovascular repair of complex aortic aneurysm involving the origin of branches, extreme angulations, complex neck anatomy, and short landing zones can be quite challenging. Use of 3D printed models can aid understanding of complex anatomy, device selection, and design of prosthesis best suited for patient’s anatomy. These models have shown to be useful in planning procedures and increase operator confidence . 3D printed models have also been used to precisely place fenestrations on stent grafts to treat complex aneurysms [479, 567]. In addition, graft replicas can be tested on patient specific 3D model for suitability before being deployed in patients.
Aortic surgeries, especially in the region of aortic arch and upper abdominal aorta can be quite challenging due to origin of branches, angulation and complex aneurysm neck anatomy. 3D printed models have shown to improve surgeons’ understanding of anatomy and help preoperative planning . Further, 3D printed models can potentially also be used to plan and simulate surgical and endovascular interventions on visceral aneurysms [502, 503]. These models can also be used for designing  and testing [568, 576] endovascular devices like catheters, coils, balloons, and stents.
Breast cancer is the most common solid malignancy in women in the United States . The overall lifetime risk of developing breast cancer for women in the United States is 12.4%. Advancements in diagnostic tests and treatments have led to decreasing death rates of 1.8% per year from 2005 to 2014 [570, 571]. Understanding the extent of disease at the time of diagnosis allows appropriate staging and determination of prognosis and survival, in addition to selection of suitable surgical options . Benefits from 3D printed models and its role as an aid to clinical care has been increasingly described in the literature. 3D printed models have the ability of depicting the extent of disease and relationships of sensitive anatomy, thereby possibly reducing operating time, enhancing utilization of new oncoplastic techniques, and improving patient outcomes.
Benign breast diseases are common and include a wide range of entities . The most common of these entities, fibrocystic change, is clinically observed in up to 50% of women and found histologically in 90% of women . Fibroadenomas are the next most common benign breast disease occurring in 15–23% of women . Surgical management of these entities may be needed in cases where cosmesis is altered or when symptom relief is needed. Surgical management may impact developing breast tissue in young women leading to alterations in its proper development . Therefore, careful understanding of the anatomy may minimize the deleterious effects of surgery in benign breast disease.
3D printing will play an increasingly important role in enabling precision medicine. This document addresses the clinical scenarios where pathology complexity necessitates a transformation of clinical imaging data into a physical model. Adoption of common clinical standards regarding appropriate use, information and material management, and quality control are needed to ensure the greatest possible clinical benefit from3D printing.
This work provides the first comprehensive literature-based guideline document regarding the implementation of 3D printing in clinical practice and details the appropriate scenarios for numerous clinical applications of 3D printing. It is anticipated that this consensus guideline document, created by the members of the RSNA 3D printing special group, will provide the initial reference for method and clinical application standardization. The document and will be substantially expanded and refined, based on expanding clinical applications.
American College of Radiology
Atrial septal defect
Computer aided design
The center for devices and radiological health
Congenital heart disease
Contrast to noise ratio
Digital imaging and communications in medicine
Double-outlet left ventricle
Double-outlet right ventricle
European Association for Cardio-Thoracic Surgery / Society of Thoracic Surgery
The United States Food and Drug Administration
Field of view
Health Insurance Portability and Accountability Act
International Classification of Diseases, Tenth Revision
International Pediatric and Congenital Cardiac Code
Magnetic resonance imaging
Not otherwise specified
Picture archiving and communication system
Partial anomalous pulmonary venous return
Region of interest
Radiological Society of North America
Right ventricular outflow tract
Special Interest Group
Signal to noise ratio
Standard tessellation language
Total anomalous pulmonary venous return
Transposition of the great arteries
Virtual reality markup language
Ventricular septal defect
Chaowu Y, Hua L, Xin S. Three-dimensional printing as an aid in Transcatheter closure of Secundum atrial septal defect with rim deficiency. Circulation. 2016;133(17):e608–10.
Faganello G, Campana C, Belgrano M, Russo G, Pozzi M, Cioffi G, Lenarda AD. Three dimensional printing of an atrial septal defect: is it multimodality imaging? Int J Cardiovasc Imaging. 2016;32(3):427–8.
Luo H, Xu Y, Wang Z, Liu Y, Gao C. Three-dimensional printing model-guided percutaneous closure of atrial septal defect. Arq Bras Cardiol. 2017;108(5):484–5.
Qiu X, Lü B, Xu N, Yan CW, Ouyang WB, Liu Y, Zhang FW, Yue ZQ, Pang KJ, Pan XB. Feasibility of device closure for multiple atrial septal defects using 3D printing and ultrasound-guided intervention technique. Zhonghua Yi Xue Za Zhi. 2017;97(16):1214–7.
Wang Z, Liu Y, Xu Y, Gao C, Chen Y, Luo H. Three-dimensional printing-guided percutaneous transcatheter closure of secundum atrial septal defect with rim deficiency: first-in-human series. Cardiol J. 2016;23(6):599–603.
Wang Z, Luo H, Gao C, Xu Y. Three-dimensional printing model for the postoperative follow-up of atrial septal defect. Int J Cardiol. 2016;222:891–2.
Yang F, Zheng H, Lyu J, Yang X, Yang Y, Pang Y, Liang F, Zhang G, Xu Z, Jiang S, Lyu B, Meng F, Hao B. A case of transcatheter closure of inferior vena cava type atrial septal defect with patent ductus arteriosus occlusion device guided by 3D printing technology. Zhonghua Xin Xue Guan Bing Za Zhi. 2015;43(7):631–3.
Bhatla P, Mosca RS, Tretter JT. Altering management decisions with gained anatomical insight from a 3D printed model of a complex ventricular septal defect. Cardiol Young. 2017;27(2):377–80.
Bhatla P, Tretter JT, Ludomirsky A, Argilla M, Latson LA, Chakravarti S, Barker PC, Yoo S-J, McElhinney DB, Wake N, Mosca RS. Utility and scope of rapid prototyping in patients with complex muscular ventricular septal defects or double-outlet right ventricle: does it Alter Management decisions? Pediatr Cardiol. 2017;38(1):103–14.
Farooqi KM, Uppu SC, Nguyen K, Srivastava S, Ko HH, Choueiter N, Wollstein A, Parness IA, Narula J, Sanz J, Nielsen JC. Application of virtual three-dimensional models for simultaneous visualization of Intracardiac anatomic relationships in double outlet right ventricle. Pediatr Cardiol. 2015;37(1):90–8.
Garekar S, Bharati A, Chokhandre M, Mali S, Trivedi B, Changela VP, Solanki N, Gaikwad S, Agarwal V. Clinical application and multidisciplinary assessment of three dimensional printing in double outlet right ventricle with remote ventricular septal defect. World J Pediatr Congenit Heart Surg. 2016;7(3):344–50.
Hadeed K, Dulac Y, Acar P. Three-dimensional printing of a complex CHD to plan surgical repair. Cardiol Young. 2016;26(7):1432–4.
Ngan EM, Rebeyka IM, Ross DB, Hirji M, Wolfaardt JF, Seelaus R, Grosvenor A, Noga ML. The rapid prototyping of anatomic models in pulmonary atresia. J Thorac Cardiovasc Surg. 2006;132(2):264–9.
Olivieri LJ, Krieger A, Loke Y-H, Nath DS, Kim PCW, Sable CA. Three-dimensional printing of intracardiac defects from three-dimensional echocardiographic images: feasibility and relative accuracy. J Am Soc Echocardiogr. 2015;28(4):392–7.
Sodian R, Weber S, Markert M, Rassoulian D, Kaczmarek I, Lueth TC, Reichart B, Daebritz S. Stereolithographic models for surgical planning in congenital heart surgery. Ann Thorac Surg. 2007;83(5):1854–7.
Valverde I, Gomez G, Gonzalez A, Suarez-Mejias C, Adsuar A, Coserria JF, Uribe S, Gomez-Cia T, Hosseinpour AR. Three-dimensional patient-specific cardiac model for surgical planning in Nikaidoh procedure. Cardiol Young. 2015;25(4):698–704.
Schievano S, Migliavacca F, Coats L, Khambadkone S, Carminati M, Wilson N, Deanfield JE, Bonhoeffer P, Taylor AM. Percutaneous pulmonary valve implantation based on rapid prototyping of right ventricular outflow tract and pulmonary trunk from MR data. Radiology. 2007;242(2):490–7.
Ryan JR, Moe TG, Richardson R, Frakes DH, Nigro JJ, Pophal S. A novel approach to neonatal Management of Tetralogy of Fallot, with pulmonary atresia, and multiple Aortopulmonary collaterals. JACC Cardiovasc Imaging. 2015;8(1):103–4.
Bhatla P, Tretter JT, Chikkabyrappa S, Chakravarti S, Mosca RS. Surgical planning for a complex double-outlet right ventricle using 3D printing. Echocardiography. 2017;34(5):802–4.
Farooqi KM, Gonzalez-Lengua C, Shenoy R, Sanz J, Nguyen K. Use of a three dimensional printed cardiac model to assess suitability for biventricular repair. World J Pediatr Congenit Heart Surg. 2016;7(3):414–6.
Farooqi KM, Nielsen JC, Uppu SC, Srivastava S, Parness IA, Sanz J, Nguyen K. Use of 3-dimensional printing to demonstrate complex intracardiac relationships in double-outlet right ventricle for surgical planning. Circ Cardiovasc Imaging. 2015;8(5):e003043. https://doi.org/10.1161/CIRCIMAGING.114.003043.
Dydynski PB, Kiper C, Kozik D, Keller BB, Austin E, Holland B. Three-dimensional reconstruction of Intracardiac anatomy using CTA and surgical planning for double outlet right ventricle: early experience at a tertiary care congenital heart center. World J Pediatr Congenit Heart Surg. 2016;7(4):467–74. https://doi.org/10.1177/2150135116651399.
Kappanayil M, Koneti NR, Kannan RR, Kottayil BP, Kumar K. Three-dimensional-printed cardiac prototypes aid surgical decision-making and preoperative planning in selected cases of complex congenital heart diseases: early experience and proof of concept in a resource-limited environment. Ann Pediatr Cardiol. 2017;10(2):117–25.
Karsenty C, Hadeed K, Acar P. Congenital heart disease: recent technical advances in three-dimensional echocardiography. Presse Med. 2017;46(5):482–9.
Biglino G, Giardini A, Hsia T-Y, Figliola R, Taylor AM, Schievano S, MOCHA Collaborative Group. Modeling single ventricle physiology: review of engineering tools to study first stage palliation of hypoplastic left heart syndrome. Front Pediatr. 2013;1:31.
Kutty S, Rathod RH, Danford DA, Celermajer DS. Role of imaging in the evaluation of single ventricle with the Fontan palliation. Heart. 2016;102(3):174–83.
Smith ML, McGuinness J, O’Reilly MK, Nolke L, Murray JG, Jones JFX. The role of 3D printing in preoperative planning for heart transplantation in complex congenital heart disease. Ir J Med Sci. 2017;186(3):753–6.
Gosnell J, Pietila T, Samuel BP, Kurup HKN, Haw MP, Vettukattil JJ. Integration of computed tomography and three-dimensional echocardiography for hybrid three-dimensional printing in congenital heart disease. J Digit Imaging. 2016;29(6):665–9.
Olivieri L, Krieger A, Chen MY, Kim P, Kanter JP. 3D heart model guides complex stent angioplasty of pulmonary venous baffle obstruction in a mustard repair of D-TGA. Int J Cardiol. 2014;172(2):e297–8.
Mottl-Link S, Hübler M, Kühne T, Rietdorf U, Krueger JJ, Schnackenburg B, De Simone R, Berger F, Juraszek A, Meinzer HP, Karck M, Hetzer R, Wolf I. Physical models aiding in complex congenital heart surgery. Ann Thorac Surg. 2007;86:273–7. https://doi.org/10.1016/j.athoracsur.2007.06.001.
Aung SC, Tan BK, Foo CL, Lee ST. Selective laser sintering: application of a rapid prototyping method in craniomaxillofacial reconstructive surgery. Ann Acad Med Singap. 1999;28(5):739–43.
Cheung LK, Wong MC, Wong LL. Refinement of facial reconstructive surgery by stereo-model planning. Ann R Australas Coll Dent Surg. 2002;16:129–32.
Schantz JT, Hutmacher DW, Lam CX, Brinkmann M, Wong KM, Lim TC, Chou N, Guldberg RE, Teoh SH. Repair of calvarial defects with customised tissue-engineered bone grafts II. Evaluation of cellular efficiency and efficacy in vivo. Tissue Eng. 2003;9 Suppl 1:S127–39.
Wagner JD, Baack B, Brown GA, Kelly J. Rapid 3-dimensional prototyping for surgical repair of maxillofacial fractures: a technical note. J Oral Maxillofac Surg. 2004;62(7):898–901.
Cao D, Yu Z, Chai G, Liu J, Mu X. Application of EH compound artificial bone material combined with computerized three-dimensional reconstruction in craniomaxillofacial surgery. J Craniofac Surg. 2010;21(2):440–3.
Levine JP, Patel A, Saadeh PB, Hirsch DL. Computer-aided design and manufacturing in craniomaxillofacial surgery: the new state of the art. J Craniofac Surg. 2012;23(1):288–93.
Cui J, Chen L, Guan X, Ye L, Wang H, Liu L. Surgical planning, three-dimensional model surgery and preshaped implants in treatment of bilateral craniomaxillofacial post-traumatic deformities. J Oral Maxillofac Surg. 2014;72(6):1138.e1–14.
Kaur H, Nanda A, Koli D, Verma M, Singh H, Bishnoi I, Pathak P, Gupta A. An alternate vista in rehabilitation of cranial defects: combining digital and manual techniques to fabricate a hybrid Cranioplast. J Craniofac Surg. 2015;26(4):1313–5.
Xie Y, Rustom LE, McDermott AM, Boerckel JD, Johnson AJ, Alleyne AG, Hoelzle DJ. Net shape fabrication of calcium phosphate scaffolds with multiple material domains. Biofabrication. 2016;8(1):015005.