The X-ray visibility of neurovascular implants like stents or coils is a crucial factor for the success of an intervention. The thin stent structures are hardly recognizable under fluoroscopy, which makes exact placement and judgment of wall apposition difficult , . For medical approval processes of such implants, radiopacity has to be proven. This is done by quantitative  or qualitative X-ray radiographies. For qualitative assessment, the implant is placed in different regions of a human skull specimen. Following the implant is imaged by radiography with clinical parameters. The visual appearance of the implant is compared to a benchmark.
Disadvantage of this method is, that the visibility strongly depends on the region of interest and the thickness and density of the used bones. Thus, images of implants from different sites are not comparable with each other.
The aim of this work was to overcome this problem by using a reproducible artificial model of the skull in a defined field of view. Following the development process of the model is described and an assessment of the model is done by comparing the radiographic data sets of the original and phantom in an implant trial.
2 Material and methods
2.1 Preparation and image generation
In cooperation with the anatomical institute, the skull of a female body donor was selected for image generation. The jawbone of the skull was removed and the calvaria was cut for preparation processes.
A cone beam computed tomography (CT) data set of the skull was acquired with a Siemens Artis zeego (Siemens Healthcare GmbH, Erlangen, Germany) flat panel angiography system (see Figure 1). The data set was 3D reconstructed with an isotropic voxel size of 0.5 mm.
2.2 Image processing
To define a realistic projection plane of the skull, neurointerventions of patients undergoing device implantation were analysed. The results were verified by experienced neuroradiologists. From the chosen projection plane, a region of interest (ROI) was selected. Structures affecting the visibility, like the temporal bone as an area with high attenuation and the opposing skullbones along the projection axis, were included in the ROI (see Figure 2). The CT dataset was processed with Mevislab (MeVis Medical Solutions AG, Bremen, Germany). A thresholding filter was used to separate bones in the ROI from patient table and imaging artifacts. The segmentation process was finished by using a region growing algorithm to segment only connected areas of the temporal bone and skullcap.
2.3 Design process
For an efficient manufacturing process the distance between skullcap and temporal bone was decreased, as changes of this distance don’t induce relevant deviations from realistic projection images (Figure 3A and B).
For connecting upper and lower bone structures in the desired orientation of the physical model, four rods were modeled in the data set (see Figure 3C).
The post processed image data was triangulated to create a standard tessellation language (STL) data type which can be printed via rapid prototyping systems. The volume model was smoothed to decrease the step effects of the voxel surface.
2.4 Rapid prototyping - techniques and materials
Requirements for the design of a realistic head phantom are similar X-ray attenuation properties as real bone and that identical structures are visible. The attenuation is described by the Hounsfield Units (HU) which vary from 50–3000 for different bones . In the skull specimen, the cortical bone had a density of ∼2300 HU, whereas the trabecular bone had lower values of ∼800 HU. The HU values were calculated from a volume data set of the human specimen, which was generated at 70 kV with a Siemens SOMATOM Definition AS+ (Siemens Healthcare GmbH, Erlangen, Germany) CT system.
Attenuation and depiction of the structures highly depend on the chosen manufacturing process and material. In a first step, the attenuation properties of different rapid prototyping (RP) materials were checked. Therefore, a cube of 1 × 1 × 1 cm3 was designed and stored as a surface triangulated file format. The cube was manufactured in different materials with the following RP machines: ProJet 6000, ZPrinter 310 Plus, Thermojet (all 3D Systems Ltd., Rock Hill, South Carolina, USA), EDEN 330 (Objet Geometries Inc, Rehovot, Israel), Freeform Pico (ASIGA Pty Ltd, St Peters, Australia), Form 1+ (Formlabs Inc., Somerville, Massachusetts, United States), LOM 2030 (Helisys Inc., Michigan, U.S.). A CT scan of each sample was performed (see Figure 4) and their arithmetic mean gray values were calculated (see Table 1). Powder material showed bone-like attenuation values, but it is limited by the manufacturing process. With the underlying 3d printing (3DP) process cavities cannot be created without remnants of powder. This increases attenuation compared to normal temporal bones, in which such cavities are filled with air. Sample 10 and 11 showed values in the range of 415 HU to 512 HU, which are less than human cortical bone but are still in the range of expected gray values. The material was chosen as a feasible compromise between imaging and manufacturing properties.
After the manufacturing process, the model was cleaned, support structures removed and fully cured in a UV light system.
3.1 Comparison of characteristic structures
The phantom was imaged with a 2D radiography and 3D volume CT. The acquired images were compared with those of the human skull specimen regarding characteristic structures (see Figure 5). Radiographic images show similarities in the gray scale characteristics, but 2D tomographic images show differences in filigree bone structures. The reason is that the cavities of the phantom temporal bone still contain liquid polymer, which has a similar attenuation as solid polymer.
3.2 Implant trial
A measurement setup was created in accordance to Schmidt et al.  based on a polymethyl methacrylate (PMMA) block of 200 mm × 200 mm × 150 mm (l × b × h) as scattering body. The bone phantom was positioned in the center of the block. A field of view of 200 mm × 200 mm was used for radiography. As usual done in clinical neurovascular interventions, the tube voltage was adjusted to 70 kV. A representative stent was positioned in the ROI. The auditory canal was used for defined alignment. A second data set of the original skull with the identical implant and orientation was acquired for a qualitative comparison. Figure 6 shows the image data with equal windowing properties (2006/1720). As expected, human bone has higher attenuation which can be determined in higher contrast values of the skull structures. In both images stent markers are distinguishable. In the phantom data set, stent struts are visible, though only barely.
The developed skull phantom showed similar attenuation distribution as a human skull. Attenuation of the chosen rapid prototyping material is lower than that of real bone. This results in lower contrast values of phantom data sets. However, no other of the tested polymeric RP materials showed attenuation properties like cortical bone. Furthermore, the varying density of bones cannot be imitated with the investigated rapid prototyping methods.
However, the presented setup can be used for generating reproducible datasets for comparison of implant visibility. Beside the qualitative testing of the radiopacity, this method can be used for generating data for medical approval processes or product benchmark.
Research funding: This work was partly funded by the Federal Ministry of Education and Research in Germany within the Research Campus STIMULATE under grant number 03FO16102A. Conflict of interest: Authors state no conflict of interest. Material and Methods: Informed consent: Informed consent has been obtained from all individuals included in this study. Ethical approval: The research related to human use complies with all the relevant national regulations, institutional policies and was performed in accordance with the tenets of the Helsinki Declaration, and has been approved by the authors’ institutional review board or equivalent committee.
Lu X, Chen T, Comaniciu D. Robust discriminative wire structure modeling with application to stent enhancement in fluoroscopy. In Computer Vision and Pattern Recognition (CVPR). 2011:1121–7. Google Scholar
Boese A, Rose G, Friebe M, Hoffmann T, Serowy S, Skalej M, et al. Increasing the visibility of thin NITINOL vascular implants. Current Directions in Biomedical Engineering. 2015;1:503–6. Google Scholar
Schmidt W, Behrens P, Kamke F, Schmitz KP. Measurement of radiopacity of vascular implants. Biomedical Engineering-Biomedizinische Technik. 2014;59:S1216–9. Google Scholar
Lehmann T, Oberschelp W, Pelikan E, Repges R. Bildverarbeitung für die Medizin: Grundlagen, Modelle, Methoden, Anwendungen. Springer-Verlag: 2013. Google Scholar
About the article
Published Online: 2016-09-30
Published in Print: 2016-09-01
Citation Information: Current Directions in Biomedical Engineering, Volume 2, Issue 1, Pages 351–354, ISSN (Online) 2364-5504, DOI: https://doi.org/10.1515/cdbme-2016-0078.
©2016 Thomas Hoffmann et al., licensee De Gruyter.. This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. BY-NC-ND 4.0