Third- and fourth-year osteopathic medical students from A.T. Still University Kirksville College of Osteopathic Medicine and residents from Northeast Regional Medical Center were invited to participate in the study. Participation was voluntary. Residency types included internal medicine, anesthesia, osteopathic manipulative medicine, and family medicine. Participants used 3D-printed medical gel joint models made in-house for mastering joint injection techniques with ultrasonography guidance. The A.T. Still University-Kirksville institutional review board granted exempt status for all educational procedures used in the current study.
Three new models were developed and used in the current study (
Figure 1). The cervical spine model included 7 cervical vertebrae with spinous and transverse processes, intervertebral discs, nerves, and vasculature. The lumbar puncture/spinal epidural model contained lumbar vertebrae L1-L5, intervertebral discs, ligamenta flava, epidural space, and subarachnoid space with cerebrospinal fluid. The pelvic model contained os coxae, lumbar vertebrae L4 and L5, intervertebral discs, nerves, sacrum, coccyx, ilium, sacroiliac joints, and posterior superior iliac spine.
Flexible intervertebral disks, nerves, and vasculature were printed using Cheetah flexible filaments (NinjaTek, Fenner Inc.) on an Afinia 3D printer. A Stratasys F123 Series: F170 printer was used to print the vertebrae, pelvis, and sacrum with acrylonitrile butadiene styrene filaments. After the models were printed and cleaned in an SCA-1200HT support removal system (Phoenix Analysis & Design Technologies), they were assembled using Gorilla glue gel and small screws. The printed nerves, vasculature, and rubber tubing (representing the spinal cord) were placed in their proper anatomical position. Some of the models were then coated with high-temperature silicone to minimize bubbles. The models were attached with screws to wooden stands to support them in the proper anatomical position.
Humimic Gel #0 (Humimic Medical Healthcare Products) is one of the most versatile gelatins used for medical imaging. The gel is clear and produces realistic tissue-like texture when imaged with ultrasonography. This gel was used to encase the models so blind insertion and ultrasonography-guided techniques could be used for injection procedures on all of the models. Molds were designed using aluminum sheet metal and flue pipe to ensure the proper shape and size for each model. The molds were attached to the wooden bases around the models with high-temperature fireplace sealant, flue tape, and steel duct clamps to ensure that the hot liquid gel did not leak. The gel was cut into small pieces and melted in 3-pound increments at 121°C. Once melted, it was poured into each mold and allowed to cool. Once the gel cooled and solidified, an electric heated knife and hot air gun were used to sculpt the molds into proper anatomical shapes.
The completed models were used in injection technique laboratories that taught needle-guidance procedures with the use of ultrasonography. Before participating in the laboratory, medical students and residents of the current study mastered basic ultrasonography techniques and applications through clinical ultrasonography coursework and monthly didactic ultrasonography sessions.
22 The laboratory began with a presentation that explained the objectives and clinical relevance of the injection exercise. The presentation was followed by a live demonstration of the injection technique and ultrasonography scanning technique using the 3D-printed models (
Figure 2). The models are also ideal for ultrasonography-guided techniques because the gel produces a realistic tissue-like texture when imaged. Participants used Mindray-5 ultrasonography machines with 10L4s linear ultrasonography transducer and a frequency bandwidth of 8.0/10.0/12.0 MHz (Shenzhen Mindray Bio-Medical Electronics Co.).
The objective of the injection ultrasonography laboratory was to identify anatomical structures of interest and perform injections. Students and residents had to identify the following anatomical structures to perform injections: C1-C7 spinous and transverse processes and facet joints on the cervical model; L1 through L5 spinous processes on the lumbar puncture model; and L4 and L5 spinous processes, sacroiliac joints, and caudal epidural region on the pelvic model.
After the training laboratory, participants were asked to complete a 17-item paper survey about their perceptions of the importance of the models for education and future practice. The survey was created specifically for the current study by the course director for the Clinical Ultrasonography course; it was voluntary, anonymous, and did not collect demographic information. The survey was validated using predoctoral fellows (completed 2 years of medical school, taking a year off to help faculty teach courses) to verify, items were readable, and interpreted consistently. Fifteen items were Likert-scale (strongly disagree, disagree, neither agree nor disagree, agree, and strongly agree) items. Those items evaluated participant comfort level with performing joint injections after using the models with ultrasonography needle guidance, overall satisfaction with the models, and likelihood of using 3D models in the future. Items also asked participants to compare commercial injection training models with the in-house 3D models (feel, anatomical correctness, and comfort when using the models). The last 2 survey items were open-ended items that asked, “What is one thing you would like to see as part of the Injection Lab regarding the use of simulation models that we did not include?,” and “Please leave comments about the use of 3D-printed models in medical education.”
The current study used a sample of convenience, so no formal sample size calculation was performed. Total scores for the 15 Likert-scale items (where strongly agree was defined as 5 and strongly disagree was defined as 1) were calculated for all participants. Survey responses were summarized with frequencies and percentages. Agree and strongly agree responses were combined and defined as suggesting agreement. Disagree and strongly disagree responses were also combined and defined as suggesting disagreement. A χ2 test was used to compare the proportion of agreement between students and residents. The binomial test of proportion was used to compare the proportions of agreement and disagreement for all individual survey items. Data analysis was completed using SAS version 9.4 (SAS Institute, Inc.). P≤.05 was considered statistically significant.