Creation of a High-Fidelity, Low-Cost, Intraosseous Line Placement Task Trainer via 3D Printing

Summary

We describe a procedure to process computed tomography (CT) scans into high-fidelity, reclaimable, and low-cost procedural task trainers. The CT scan identification processes, export, segmentation, modeling, and 3D printing are all described, along with the issues and lessons learned in the process.

Abstract

The description of procedural task trainers includes their use as a training tool to hone technical skills through repetition and rehearsal of procedures in a safe environment before ultimately performing the procedure on a patient. Many procedural task trainers available to date suffer from several drawbacks, including unrealistic anatomy and the tendency to develop user-created 'landmarks' after the trainer tissue undergoes repeated manipulations, potentially leading to inappropriate psychomotor skill development. To ameliorate these drawbacks, a process was created to produce a high-fidelity procedural task trainer, created from anatomy obtained from computed tomography (CT) scans, that utilize ubiquitous three-dimensional (3D) printing technology and off-the-shelf commodity supplies.

This method includes creating a 3D printed tissue mold capturing the tissue structure surrounding the skeletal element of interest to encase the bony skeletal structure suspended within the tissue, which is also 3D printed. A tissue medium mixture, which approximates tissue in both high-fidelity geometry and tissue density, is then poured into a mold and allowed to set. After a task trainer has been used to practice a procedure, such as intraosseous line placement, the tissue media, molds, and bones are reclaimable and may be reused to create a fresh task trainer, free of puncture sites and manipulation defects, for use in subsequent training sessions.

Introduction

Patient care competency of procedural skills is a critical component for developing trainees in civilian and military healthcare^1,2 environments. Procedural skills development is particularly important for procedure-intensive specialties such as anesthesiology³ and front-line medical personnel. Task trainers may be used to rehearse numerous procedures with skill levels ranging from those of a first-year medical student or medical technician to a senior resident or fellow. While many medical procedures require significant training to complete, the task presented here-placement of an interosseous (IO) line-is straightforward and requires less technical skill. Successful placement of an IO line can be accomplished after a relatively short period of training. The use of simulation during medical training, which includes the use of task trainers, is recognized as a tool to gain technical procedural skills through the repetition and the rehearsal of a clinical procedure in a safe, low-stress environment, before ultimately performing the procedure on patients^2,4,5.

Understandably so, simulation training in medical education environments has become widely accepted and appears to be a mainstay, despite the paucity of data regarding any impact on patient outcomes^6,7. Additionally, recent publications demonstrate that simulation improves team performance and patient outcomes as the result of improved team dynamics and decision-making. Still, there is little data to suggest that simulation improves the time or success rate to perform critical, life-saving procedures^8,9 suggesting that simulation is complex and multifaceted in the education of health care providers. In patients where standard intravenous access is not possible or indicated, IO line placement may be used to achieve vascular access quickly, requiring minimal skill. Timely and successful performance of this procedure is critical, particularly in the perioperative environment or a trauma scenario^10,11,12. Because IO line placement is an infrequently performed procedure in the perioperative area and can be a life-saving procedure, training in a non-clinical environment is critical. An anatomically accurate task trainer specific to IO line placement is an ideal tool for offering predictable training frequency and skills development for this procedure.

Although widely used, currently available commercial task trainers suffer from several significant drawbacks. First, task trainers that allow for multiple attempts of a procedure are costly, not only for the initial purchase of the task trainer but also for replenishing the replaceable parts such as silicone skin patches. The result is often infrequently replaced parts, leaving prominent landmarks that provide the trainee a suboptimal training experience; patients will not come pre-marked where one should do the procedure. Another drawback is that the high cost of traditional task trainers can result in limited access by users when the devices are 'locked up' in protected storage locations to prevent loss or damage to the devices. The result is requiring more rigorously and less available scheduled practice time, limiting their use can certainly make unscheduled training difficult. Finally, most trainers are considered low-fidelity^5,13,14 and use only representative anatomy, potentially leading to inappropriate psychomotor skills development or training scars. Low-fidelity trainers also make the thorough assessment of skill acquisition, mastery, and degradation very difficult as training on a low-fidelity device may not adequately mimic the actual real-world procedure.

Representative anatomy also impedes the proper evaluation of the acquisition and mastery of psychomotor skills. Moreover, assessing the transfer of psychomotor skills between simulated medical environments to patient care becomes nearly impossible if some of the psychomotor skills are not reflected in the clinical task. This results in the prevention of consensus on the ability of medical simulation and training to affect patient outcomes. To overcome the challenges of cost, anatomical accuracy, and access, we have developed a low-cost, high-fidelity IO line task trainer. The task trainer is designed from an actual patient's CT scan, resulting in accurate anatomy (Figure 1). The materials used are ubiquitous and easy to obtain, with components that are relatively easy to reclaim. Compared to many other commercially available trainers, the modest cost of the task trainer design described here dramatically reduces the desire to sequester the trainers in a less accessible, protected location and makes multiple repetitions without leading landmarks possible.

Protocol

NOTE: The University of Nebraska Medical Center Institutional Review Board determined that our study did not constitute human subject research. The local IRB obtained ethical approval and waiver of informed consent. Complete anonymization of imaging data was done before analysis per the hospital de-identification protocol.

1. Data

1. Obtain a CT scan capturing the anatomy of interest for the planned task trainer. Be careful to take into consideration the working volume limitations of the 3D printer used and required landmarks for procedural steps.
2. If the scan is obtained in a Digital Imaging and Communications in Medicine format (DICOM), convert to a Neuroimaging Informatics Technology Initiative (NiFTi)¹⁵ format (.nii).

2. Segmentation

1. Use 3D Slicer software (http://www.slicer.org) to segment the CT images. Import the NIfTi file from Step 1.2 into 3D Slicer.
2. Select the Segment Editor module to generate the segments needed to model the trainer.
   1. Add one segment for the 1) Bone and 2) Tissue components of the task trainer.
      NOTE: Development of some trainers, such as those used to train chest tube insertion, may require additional segments.
   2. Select segment 1) Bone. Using the Threshold Effect, change the intensity range until the defined "window" range identifies the Bone component of interest.
      NOTE: For bone segments the usual range is between 100 and 175 HU (Hounsfield Units) to the available maximum value and for tissue, which is typically -256 HU to the available maximum.
   3. Use the Threshold function to highlight the 1) Bone component and apply it to the scan using the Apply command.
   4. Use the Scissors function to remove any areas of the scan not needed to create the task trainer. Use care to ensure that the bone marrow space remains hollow for IO trainers.
      NOTE: This step is the first reduction of the segment of interest to the desired dimensions of the trainer. The build volume limitations of the 3D printer to be used should be considered here; however the segment may be further reduced in section 3.
3. Repeat steps 2.2.1-2.2.4 for the 2) Tissue component.
4. Using the Segmentations module; export each component as an STL file.

3. 3D Modeling

1. Use AutoDesk Meshmixer to crop the 3D segments further and reduce the resolution of each segment, in terms of the number of geometric elements, for optimal performance within Fusion360.
   1. Confirm that imported STL files have the correct triangle normal orientation. Ensure the normals of each triangle point in the direction of the outer surface of the mesh. If the triangle orientation is incorrect, flip the triangle normal by performing the Select | Modify | Select All function and then the Select | Edit | Flip Normals function.
   2. Eliminate unwanted structures (e.g., unwanted segments of tissue or vasculature captured by the CT due to the use of contrast) of the imported STL Segments, and refine the models needed to create the task trainer. To refine the model by removing unwanted structures within the segments that may have been inadvertently included within the threshold range of the exported segment, use the Select operation, select the triangles on the undesired structures, then Edit | Discard.
   3. Following 3.1.2, use the Edit | Plane Cut tool to crop the model to fit within the confines of the 3D printer build volume. To reduce the computational overhead incurred due to excessive geometric resolution, reduce the number of triangles used to define the model to allow for optimal performance in Fusion360. Click on Select, double-click anywhere on the mesh to select the entire mesh, then Edit | Reduce. For Reduce Target, reduce to a Triangle Budget of under approximately 10,000 faces.
      NOTE: The printer currently used by the authors has a maximum build volume of 250 x 210 x 210 mm; thus the model was cut to a maximum long-axis length of 220-230 mm to allow the mold to fit within the printer's build volume. The printer's build volume should dictate the long-axis length by making the model approximately 20-30 mm shorter. The geometry can easily be reduced to ~10K triangles without loss of clinically relevant detail to develop high-fidelity task trainers.
   4. Eliminate or reduce holes and surface irregularities using the Select tool. Once the triangles of the mesh around the defect are selected, use the command Select | Edit| Erase&Fill to improve surface holes and irregularities.Export and save the finished models using the STL file type.
      NOTE: The outer surface of the target bone for the interosseous line task trainers requires complete closing; otherwise, the melted tissue media will enter the marrow space and degrade the task trainer performance.
2. Use AutoDesk Fusion360, and import the bone and tissue models by adding the .STL files into the workspace as a mesh using the Insert | Insert Mesh command.
   1. Convert the imported meshes into BRep solids by disabling the Fusion360 timeline and reducing the number of triangles in the target mesh to <10,000.Select the imported Mesh Body and right-click. Choose the Mesh to BRep option. After the meshes have been converted to BReps solids, resume the Fusion360 timeline.
   2. Modify the solid to create the Task Trainer's mold by splitting the rectangular solid along the long axis of the Tissue BRep.
      NOTE: The mold is created around the Tissue BRep by using the sketch feature to build a cube or rectangular solid that encompasses the Tissue solid. The mold size should be modified to meet the maximum build volume of the selected 3D printer. As the mold is split in two, the longest dimension printed may not be the final mold's largest dimension as they are joined.
   3. Select 2-3 locations for support pins, and place the pre-designed assembly group components to fix the task trainer's bones. Make sure that the locations selected for the support pins have an ample support structure in the bone around the head of the pin.
      NOTE: The bone around the pin head selected does not need to be perfectly uniform as the assembly group also contains a solid cylindrical support structure, which will be fused with the bone. This structure adequately supports the head of the pin and preserves correct anatomic placement of the bones within the tissue media.
   4. Import and position a bone plug onto the open marrow space of the Bone BRep to prevent tissue media from entering the marrow space, and keep the simulated bone marrow from draining out.
   5. Generate an opening (typically 4-6 cm in diameter) through the molds in the space represented by the Tissue BRep solid to permit pouring the liquid tissue media into the mold.
   6. Once the components of the pre-designed assembly groups are positioned to fix the bones in space, perform Boolean Combine functions to either add or cut the various assembly groups into the models.
      1. Perform a mirror of the objects before step 3.2.6 to make the task trainer for the ipsilateral side. Repeat steps 3.2.3-3.2.5 before 3.2.6.
   7. Export the final components for printing. Select the desired body within the workspace and generate an STL file via right-click | Save As STL.

4. 3D Printing

1. Using Simplify 3D, position the STL file on the bed of the 3D printer so that the slicing program may generate the GCODE required to print the item. Print the components with Poly-lactic Acid (PLA) 3D printer media filament using a 0.4 mm nozzle at a hot end temperature of 210 °C. Make sure that the settings utilize 4 top and bottom layers and 3 perimeter shells.
2. Orient the bones vertically to minimize the required support material within the marrow cavity. Print using a raft, 0.2 mm layer height, 20% infill, and full support material (from the print bed and within the print). When printing the tissue molds, orient the mold components with the tissue surface facing up. Print the tissue molds without a raft, 0.3 mm layer height, 15% infill, and full support material.
3. Arrange the support pins and other components to minimize support material-print all pin support parts with a raft, 0.2 mm layer height, and 20% infill. Print the threaded components without support material at a reduced speed, to maximize the fidelity of the thread structures.
4. Once each component's parameters are selected, prepare and export the GCODE file generated by Simplify 3D to an SD card.Using a Prusa i3 MK3, select the saved GCODE file from the SD card and print with 1.75mm PLA 3D printer media filament.

5. Assembly

1. Prepare the tissue medium.
   NOTE: The trainee's current level of skill mastery may dictate whether opaque or transparent tissue medium is required. Transparent medium allows the trainee to visually track their progress during IO insertion and more easily identify bony landmarks, while opaque medium better simulates actual clinical experience.
   1. Measure the following components to be used to create the tissue media, and set aside (these quantities may be scaled as needed) 260 g of unflavored gelatin; if required, 140 g of finely ground psyllium husk fiber, orange-flavored, sugar-free (omit this step to create a transparent medium); 42 g of 4% w/v chlorhexidine.
      NOTE: Psyllium husk fiber may be used to make an opaque medium. This component should be added immediately after the gelatin if an opaque medium is desired¹⁶.
   2. Heat 1000 mL of water (tap is acceptable) to 85 °C.Add the water to a mixing container several times larger than the volume of ingredients, such as an 18.9 L bucket.
      1. While vigorously mixing the tissue medium solution, add the gelatin, psyllium husk fiber, and chlorhexidine solution to the water, in order, and wait before adding the next ingredient after the previous one is incorporated.
         NOTE: Do not add psyllium husk fiber if making transparent medium.
   3. Heat the mixture in a 71 °C water bath for a minimum of 4 h to allow the bubbles to dissipate from the solution. Place the mixing container in the hot water bath directly, or transfer the mixture to a separate container, such as plastic storage bags.
   4. Prepare the tissue medium for pouring into the assembled mold. Ensure that the mixture is homogeneous and fluid. Maintain the temperature of the mixture at 46 °C.
      NOTE: If the tissue medium is not immediately needed, it may be stored at 4 °C or -20 °C within a storage container until needed.
2. Prepare the simulated bone marrow solution.
   NOTE: The simulated bone marrow solution may be prepared in advance and stored in a covered container at room temperature until ready for use.
   1. Measure and thoroughly mix 100 g of cool water (tap is fine); 100 g of ultrasound gel; and 5 mL of red food coloring (optional, used to improve simulation). Ensure that the final product is thick but fluid enough to transfer quickly.
3. Secure the bone to the bottom of the mold, and assemble the mold.
   1. Spray each side of the mold's inner surfaces with a non-silicone-based releasing agent, such as non-stick cooking spray.Secure the bone using the support pins to maintain the correct position within the tissue space. Secure the bones/pins to the bottom of the mold.
   2. Align the top of the mold to the bottom portion, and secure the two halves of the mold together. Verify the bone plug is in position to prevent tissue medium entering the marrow space during pouring.
4. Position the mold such that the opening is facing up, and pour the 46 °C tissue medium into the mold's cavity.Remedy any leakage of the tissue medium from the mold using an inverted air duster canister by directly spraying the warm tissue medium with the canister to cool it quickly.Transfer the filled mold to a 4 °C refrigerator for a minimum of 6 h, or until the tissue medium has set.
5. Disassemble the mold, and remove the task trainer and the support pins.Remove the bone plug, fill the marrow space with simulated 'bone marrow' created in 5.2, and replace the bone plug. Place the task trainers in a plastic storage bag, and store the assembly at either 4 °C or -20 °C until needed for training.

6. Task training

1. Remove the task trainer from storage and allow it to reach room temperature. If not already in place, add simulated bone marrow material from step 5.2 per instruction in 5.5.
   NOTE: Allowing the trainer to warm to room temperature improves the simulation experience.
2. Perform training on the task trainers. Instruct the trainees to place IO needles (Figure 2A), and aspirate simulated bone marrow (Figure 2B) as per the IO line placement's usual steps.
3. Following training, disassemble the task trainers to reclaim tissue, the medium, and the bones.
   NOTE: After manipulation, the bones of the IO trainer will have holes created by insertion of the IO line canula. These holes may be either filled with PLA using a handheld 3D printer pen, or alternately the bones may be discarded.
4. Reassemble and reuse reclaimed materials for subsequent training as per section 5.Alternatively, melt the tissue medium down, reclaim per 5.1.4, and store at either 4 °C or -20 °C, if not immediately needed.

Results

Following the protocol, the modeling of the task trainer utilized a CT scan of a de-identified patient. Segmentation of the CT images utilized 3D Slicer software and Auto Meshmixer for 3D modeling. For 3D printing, both 3D Simplify and the Prusa i3 MK3 were used (Figure 1). Subsequently, we completed the assembly of the 3D-printed parts, prepared the tissue media mixture, and poured the media mixture into the assembled task trainer mold. Following a training period with the task trainer, the...

Discussion

In this protocol we detail a 3D task trainer's development process to train the infrequently performed and life-saving procedure of IO line placement. This self-guided protocol uses 3D printing to produce the bulk of the model structures, while the remainder of the components used to assemble the task trainer are ubiquitous, easily obtainable, and non-toxic materials that may be reclaimed and reused. The 3D task trainer is low-cost and requires minimum expertise to create and assemble. We have successfully used our 3D IO...

Materials

Name	Company	Catalog Number	Comments
3D printer filament, poly-lactic acid (PLA), 1.75 mm	N/A / Hatchbox		Base for 3D printing molds, bone structures, and bone / mold hardware
3D printer, Original Prusa i3 MK3	Prusa		To print molds, bone structures, and bone / mold hardware
bolts, 1/4”, flat / countersunk or round head, various lengths	N/A		Hardware used to hold mold casing halves together during casting
Bucket, 5 gallon, plastic	N/A		To hold tissue media during media preparation
chlorhexidine, 4% solution w/v			Animicrobial additive for tissue media
drill, household 3/8’ chuck	N/A		To stir tissue media during media preparation
food coloring, red (optional)	N/A		Coloring additive for simulated bone marrow
gelatin, unflavored	Knox		Base for tissue media
hex nuts, 1/4”	N/A		Hardware used to hold mold casing halves together during casting
Non-stick cooking spray	N/A		Mold releasing agent
plastic bags, ziplock	Ziplock		To store tissue media
psyllium husk fiber, finely ground, orange flavored, sugar free (optional)	Procter & Gamble	Metamucil	Opacity / Echogenicity additive for tissue media
screwdriver, flat / Phillips (matching bolt hardware)	N/A		To tighten mold casing hardware
silicone gasket cord stock, 3 mm, round, various lengths	N/A		Gasket media for mold casings
spray adhesive, Super 77 (optional)	3M		Agent used to improve bed adhesion during 3D printing
stirring paddle / rod			To stir tissue media during media preparation
turkey baster, household, 60 mL	N/A		To inject simulated bone marrow into bone marrow cavity
ultrasound gel			Base for simulated bone marrow
water, tap			Used in both tissue media and simulated bone marrow