There are different factors that have to be considered when implementing a 3D printer in a vascular surgical environment, but from our limited experience and from literature^2,3, the most important ones are as follows:

Technology used

Fused deposition modeling ( FDM ) or stereolithography ( SLA ) are the most common technologies used, both with their strengths and weaknesses^4,5. So far we have worked only with an FDM printer and we’re planning to add a resin printer ( SLA ) to our setup.

Table 1: Comparison between FDM and SLA 3d printing technologies.

Sizing and resolution:

Taking into account the size of the build platform is essential for a seamless 3D printing process without having to separate the model, which comes with prolonged printing time, further model editing and different bonding methods for the resulting parts.

Resolution of the 3D printer plays an important aspect particularly in small lumen vessels, for the aorta is of less importance⁶.

Materials

Of all the materials available for commercial 3D printing, finding a suitable one for reproducing a vascular model is by no means an easy task⁷.

Table 2: Required characteristics of materials used in 3d printed vascular models .

Table 3: Materials we have tested so far. Both are recyclable.

Fig 1: Technical datasheet with material properties provided by the manufacturers.

 

Price

The price range of 3D printers is rather large, varying from Desktop machines to Industrial printers but that is only the initial investment, the cost of materials and maintenance should also be considered in the long term.

Table 4: Characteristics of the printer used -...

• Process

We propose a step by step approach of the printing process of aortic models, starting with image acquisition, patient selection, image post-processing, 3D printing and ending with printed object post-processing and preoperative planning.

The steps described below, apart from printed object post-processing, are similar to all types of 3D printers regardless of the technology or the materials used in the procedure.

• Image acquisition

CT angiography (CTA) is currently the gold-standard modality of aorta imaging⁸. Images obtained through this method are saved and stored in Digital Image and Communication in Medicine (DICOM) format.

For a smooth 3D printing process the images procured by CTA should have no image artifacts or as low as possible, optimal contrast for the region of interest, isotropic voxel resolution and low noise. An image thickness between 0.5-1.5 mm is desirable.

• Patient selection

As a consequence of the time and resources needed for 3D printing aortic models, patients that require elective surgery are better suited for this type of medical solution.

• Postprocessing of DICOM images

Since this format (DICOM) does not allow us to directly print, we need to separate the structures of interest from the raw data and transform it into a format accessible to 3D printers, the most notable one being STL (stereolithography). The main tool we utilized for segmentation is 3D Slicer⁹, a free, open source, cross-platform software.

After exporting data from the PACS environment, preferably anonymised, we import the data set to 3D Slicer and extract a reconstruction of the vascular structure of interest consisting of intraluminal contrast. This is called a segmentation ( isolation of anatomical structures from their surroundings).

There are various tools useful in achieving this task, either manually or using certain semi automated techniques (thresholding, edge detection, region segmentation, atlas based) or automated techniques (AI, supervised or unsupervised) or a mix of the above mentioned¹⁰.

All the models were made using a combination of semiautomatic techniques and then manual editing.

• 3D Model adjusting

The STL data provided by 3D slicer is further enhanced utilizing Meshmixer ( @ Autodesk Inc.) by smoothing the model and subtracting the luminal volume to provide a hollow model. There are a few aspects that need to be considered in post-processing of the 3D structure before printing:

1. eliminating non-manifold surfaces or edges;
2. repairing the mesh by filling unwanted holes;
3. removing or reconnecting mesh islands to the main mesh;
4. ensuring mesh quality so that printing resolution correlates to image resolution;
5. remeshing and eliminating unnecessary triangles from the mesh;
6. selecting the desired wall thickness.

All the models have been remeshed ( reducing the mesh varying between 20-30% ), for a lower triangle count and faster processing, but preserving shape and size ( life sized models ) , smoothed out ( 10% factor ), then fixed using the Analysis tool, ensuring the mesh quality with the Flat fill function of various defects listed above.

The wall thickness was arbitrarily set at 0.4 mm, respectively 1 mm while hollowing the model, and the ends were discarded using the plane cut method.

• Printing

All the prints were done on the Prusa Mini at 0.20 mm quality ( layer height ) as to be less time consuming but still retain a decent aspect, with variable layer height ( 0.1-0.2 mm ) on areas with more details or difficult overhangs and curvature.

The printing speed and extruding temperatures used were the recommended ones for the specific material. Infill values between 0-15% were used having a more transparent finish for lower values but less resistance and structural integrity. Supports were automatically generated and manually edited to assure a smooth printing process.

• Post processing of 3D objects

After cooling the print supports were gently removed using different cutting tools and pliers. Some internal parts of the supports could not be removed entirely because of difficult access. No sanding or physical smoothing was done so as to not reduce the transparency of the prints.

Salt annealing (transformation of amorphous polymers into the semi-crystalline structure) was tried using a heat convection oven at a temperature of 110° and 150°C for 30 to 45 minutes with little success in achieving better transparency and the resulting deformities being rather significant¹¹.

• Case 1.

Pseudoaneurysm of the aortic cross in a patient with multiple pulmonary hydatid cysts. The model was used before endovascular repair consisting of pseudoaneurysm removal and stent grafting using a Gelweave Graft 8 mm.

Fig 2: Saccular aortic pseudoaneurysm involving the aortic cross with dimensions of...

Fig 3: The model allowed clear understanding of the neck of the pseudoaneurysm and...

• Case 2. 

Ascending aortic aneurysm involving the aortic root and sinotubular junction. Images were obtained using a ECG gated CTA, at 68% R-R interval to avoid pulsation artifacts. The patient underwent an aortic valvular replacement ( biological valve Medtronic Hancock II nr. 25 ), non-coronary sinus plasty and aortic replacement with vascular prosthesis Gelweave nr. 34.

Fig 4: The model was used to provide an initial understanding of the actual size of...

• Case 3. 

Saccular aortic aneurysm involving the descending thoracic aorta. Endovascular aneurysm exclusion ( TEVAR technique ) and stent graft (VALIANT Captivia) placement were practiced.

Fig 5: Saccular aortic aneurysm measuring 1,6/2,1/2,8 cm AP, TR and CC with minimal...

• Case 4.

Infrarenal aortic aneurysm with associated dissection and partial thrombosis of the true lumen;  incidental finding - aberrant left renal artery. Aortic interposition graft placement and aneurysm removal with IMA ( inferior mesenteric artery ) reimplantation were practiced.

Fig 6: Fusiform aneurysm measuring about 7/7/10 cm AP/TR/CC with mural thrombus...

Fig 7: Total printing time was 8:30h for rPET.

• Results

Utilizing an open-source software setup and a desktop 3D printer allowed us to produce vascular models capable of being important teaching tools regarding endovascular/surgical procedures in aortic aneurysms, pseudoaneurysms or dissections.

The choice of procedure, stents and grafts utilized was made using the CT images, but the models were used after to better assess the anatomy and to build confidence in the selected surgical procedure¹².

Regarding the transparency of the models perhaps an SLA printer will have better results, the models obtained being at best translucent at the lowest values of infill that still provided structural integrity. Also SLA technology would reduce part of the internal supports that are difficult to remove.

Although flexible models were obtained, none were transparent enough or watertight. Nonetheless this setup can be a minimal cost stepping stone into 3D printing aortic wall models.

Fig 8: Showcasing TPU prints with a shore hardness of 90A. Despite being UV light...