Assessment of the imaging properties of 3D-printable material using dual energy computed tomography
Abstract
Aim: Assessment of the imaging properties of 3D-printable materials using dual energy computed tomography (DECT) to match clinical values for imaging phantoms. Methods: 3D-printed samples were imaged using DECT. Regions of interest were analyzed to assess spectral computed tomography (CT) numbers at various energies and measure the electron density (ρe) and effective atomic number (Zeff). Results: Electron density was proportional to the CT number for the materials assessed with Zeff between 6.43 and 7.01. The measured CT number increased with monochromatic energy for all but one sample. Conclusion: A single DECT scan provides valuable information regarding the properties of 3D-printable material due to the ease of measurement of ρe and Zeff. The majority of 3D-printed materials analyzed behaved like adipose tissue across a range of energies in CT imaging.
Plain language summary
To investigate how different 3D-printed materials respond when they are scanned using a specialist computed tomography (CT) imaging system called dual energy CT. The response of the materials can be matched to that of human tissue to produce realistic substitutes for testing scanners. 3D-printed samples were imaged using DECT. These were analyzed to assess the material response at different imaging energies and to measure the electron density (ρe) and effective atomic number (Zeff) of the materials. All but one of the 3D-printed materials analyzed showed a response similar to fat. The DECT technology provided a quick and easy measurement of ρe and Zeff. A single DECT scan provides valuable imaging information regarding the properties of 3D-printable material, compared with standard CT. The majority of 3D-printed materials assessed are not suitable for representing human tissue.
3D-printed technology was invented in the 1980s, and by the mid-1990s, experimental uses in the medical field were under development [1]. In the past 10 years as the technology has advanced, 3D printing has been applied in various applications in the healthcare sector – for example, in surgical practice for customized implants, prosthetics and medical modeling [2,3]. Many centers have also begun to investigate the potential for 3D-printing test objects, or phantoms, for diagnostic imaging [4–9]. Commercially available phantoms are often expensive, whereas 3D printing is generally cheaper [4] and enables custom design, personalized to local equipment and patients. However, when imaging with ionizing radiation, the radiological properties of the material are critical. This work aims to assess the imaging properties of 3D-printed materials for use as tissue mimics for ionizing radiation imaging phantoms. The properties of these materials are evaluated using dual energy computed tomography (DECT) and compared with other printable materials assessed in the literature. DECT is a relatively new technique that can provide more information about the properties of materials than single energy CT (SECT).
Medical phantoms
Previous generations of phantoms for diagnostic imaging used materials such as epoxy resin or polyurethanes to mimic tissue [8,10]. Most 3D-printer manufacturers offer a wide selection of materials with precise properties depending on the manufacturer and the compatibility of the printer. Fused deposition modeling, stereolithography and PolyJet prints are typically used for medical applications [11]. The size and resolution of the print is dependent on the printer used and may influence phantom design, depending on what technology is available. When using a fused deposition modeling printer, Dancewicz et al. [12] were able to print with different infill densities, leaving air gaps in the printed material. This allowed for lower density equivalent material suitable for mimicking lung tissue. The support material available for the printer is also a consideration; small structures (under a millimeter in thickness) will generally not withstand the removal of support material and the postprinting cleaning process [13].
Materials for printing phantoms for diagnostic image quality assurance require particular qualities, as summarized in Table 1. Additionally, any phantom to be used as a fillable phantom for nuclear medicine applications must be waterproof to prevent radioactive contamination.
Property of the material | How the property is measured | Effect on the final phantom |
---|---|---|
Cost | Price | Affordability |
Print resolution | Printer resolution | Affordability/accuracy |
Size of print | Printer print area | Affordability |
Waterproofness | Leakage | Practicality |
Water absorption | % absorbed | Deformation |
Colour/opaqueness | Visual | Cosmetic |
Durability | Length of working life | Single use/multiuse |
Strength | Ability to withstand use for which it is intended | Working life |
Flexibility | Shore Hardness | Movement/motion of the phantom walls |
Weight | Maximum weight for lifting/moving | Practicality – is it possible to move it between scanners? Will water be added before use? |
Imaging properties | Density, effective atomic number, electron density, mass attenuation coefficient, computed tomography number | Does it match tissue being simulated? |
If the material is susceptible to water absorption, this could lead to deformation of the phantom [14] and introduce a potential radiation risk in phantoms used for nuclear medicine. If the phantom is to be transported, the overall weight of the final phantom, including any filling if required, should be considered for practical purposes. Use of transparent material will assist in any filling processes, and materials may be colored to distinguish between different structures [15]. Properties required for a specific phantom, such as flexibility of the material, durability and strength, depend on what the test is trying to achieve or quantify. The shore hardness value is a measure of how soft or hard a material is [16] and is provided by the printer manufacturer. There are 12 standards for shore hardness, the most common two being Shore A (soft) and Shore D (hard). The standards themselves are rated on a scale of 0 to 100, with 100 being the hardest [16]. Depending on the printer, materials can be mixed during the printing process to produce different shore hardness.
Radiation properties for imaging
Imaging properties are not generally provided by the manufacturer; however there have been numerous studies assessing these properties [2,3,5–9,12–14,17–25], with many different parameters analysed to provide an estimate of tissue equivalence, including mass density (ρ), or effective atomic number (Zeff). The mass attenuation coefficient [26], defined in Equation (1), is also widely quoted, and depends on the component elements of a material. The mass attenuation coefficient of a material at a given energy can also be described in terms of the photoelectric and Compton scatter (including Coherent scatter), the contributions of each being dependent on the material atomic density and the electron density (ρe) [27].
where μ = linear attenuation coefficient [cm-1], ρ = mass density [g/cm3], and σtot = is the total atomic cross section [cm2/atom], mu = atomic mass unit [g] and A = relative atomic mass [28].
Materials commonly used for tissue equivalent phantoms include nylon, polylactic acid (PLA) and acrylonitrile butadiene styrene (ABS). Details of their composition have been published [9,10,29]; however, the elemental composition of 3D-printable material is not always available due to commercial sensitivity [14].
The linear attenuation coefficient can be calculated using the CT number, in Hounsfield units [HU], defined in Equation (2) [30], measured directly using CT imaging. The mass attenuation coefficient can then be determined using Equation (1), when the density of the material is known. However, the linear attenuation coefficient (μ), and therefore CT number, are known to differ with kilovoltage (kV) [30]. Therefore, to characterize fully the radiation properties of the material, linear attenuation coefficients would need to be ascertained across the spectrum of radiation used clinically.
The CT number is the most widely quoted parameter for assessment of the radiological properties of 3D-printed materials [4]. When assessing the imaging properties of 3D-printed materials, samples are typically printed, and a measurement of density, CT number, or the attenuation coefficient of the samples is performed. These parameters are then matched to the organ of interest. PLA- and ABS-printable materials are generally regarded as a good tissue mimics over the energy ranges tested [21].
CT data can also be used to obtain imaging information at energies commonly used in nuclear medicine. Gear et al. [14] measured the CT numbers of printed materials and used the NIST X-COM software [31] to estimate the material attenuation of each material by approximation of Zeff. From this, the linear attenuation at different energies commonly used in nuclear medicine was obtained. Alssabbagh et al. [20] used electron microscopy to assess the elemental composition by mass of the material, which then allowed for the calculation of the mass attenuation coefficient using the NIST X-COM software [31] for the energy range of iodine-131.
Although the CT number and ρ of a material can be measured with relative ease and provide an estimate of tissue equivalence, ρe and mass energy-absorption coefficient (μen/ρ) provide further information about the radiological properties of the material and are preferential parameters to use when measuring tissue equivalence for ionizing radiation [32], particularly for radiotherapy purposes in treatment planning [33]. The CT number, although proportional to ρe, changes with energy. As X-ray energy increases, the linear attenuation, and hence CT number, generally decrease. An exception to this is fat [30], which has an Zeff and ρe lower than that of water [34].
DECT allows for measurement of ρe, Zeff and virtual monochromatic CT numbers at different keV [30]. This is achieved by measuring the attenuation coefficient of the material at two energies [27]. This method has been demonstrated to measure Zeff and ρe accurately, across a range of imaging systems [34–37] and allows for characterization of materials over the full spectrum in a single scan. Additionally, measurement of the Zeff allows for accurate modeling of the material using Monte Carlo techniques, which may be used for phantom development.
Study purpose
A literature search was performed to collate results in published data that assessed the CT imaging properties of different 3D-printable materials.
DECT was then used as a quick and novel way to assess the tissue equivalence of different 3D-printable materials, including myocardial and vascular materials recently introduced to the market by Stratasys® for their J750 Digital anatomy™ 3D printer [38]; sample materials for the Stratasys Connex3 Object 350 printer [39]; materials printed on the Form 2 printer (Formlabs [40]). The CT numbers obtained by DECT and SECT at different energies was also measured for the materials and compared with the results found in the literature and those found clinically.
The imaging properties of different printable materials may be matched to other required properties, such as transparency and flexibility, to find a suitable tissue equivalent for imaging phantoms.
Materials & methods
Review of published data
A literature search was performed to identify studies where the CT imaging properties of 3D-printable material had been assessed. Inclusion criteria were as follows: published study, publicly available (e.g., the manufacturers specification), CT imaging and 3D-printable material. The literature search was performed in June 2022 using EMBASE (from 1974), OVID Medline (from 1946), all EMB reviews (including Cochrane) and OVID Medline (ahead of print). The following search terms were applied;
1. | Computed tomography OR CT | ||||
2. | 3D printed OR 3D printer OR 3D printing OR 3D print | ||||
3. | Phantom OR test object | ||||
4. | 1 AND 2 AND 3 |
References from the included papers were searched to snowball further studies. Selection was by the author as per the criteria listed above. After duplicate removal, the abstracts were screened first, followed by the full text of the papers, with those not meeting the criteria excluded. The process is shown in Supplementary Figure 1.
The materials, printer, density of the material used, the CT number (HU) and the measured standard deviation, where available, at different kV were collated from the literature.
Assessment of materials with DECT
Three sets of printable materials were tested. Four samples (cubic; 1 cm × 1 cm × 1 cm) of the coronary artery, carotid artery, aorta and five myocardium materials (myocardium 1, 2, 3, 4 and 5) were printed using the Stratasys J750 Digital Anatomy 3D Printer [38] (n = 8 sets) provided by Aiju (Instituto Technológico de Producto Infantil y Ocio, Alicante, Spain) [41]. The material properties of the samples, such as flexibility, have been previously investigated [42,43]. A pre-printed sample set of PolyJet Digital Materials (cuboid with rounded edges; 8 cm × 2.5 cm × 0.4 cm) provided by Stratasys and compatible with the Stratasys Connex3 Object 350 printer [39], available at our institution, were also imaged (n = 18). The scan direction was along the 2.5 cm axis. Physical properties of these materials are published online by Stratasys in their materials datasheet [44]. Finally, three samples (cubic; 2 cm × 2 cm × 2 cm) were printed on the Form 2 printer (Formlabs [40]), also available at our institution.
The samples were lined up along the isocenter and scanned in air using a Siemens SOMATOM Definition Flash Dual Energy CT scanner (Erlangen, Germany), using both DECT and SECT, with the settings in Table 2.
kV | mA | Rotation time (s) | FoV (mm) | Pitch | Recon algorithm | Recon thickness (mm) | Collimation thickness (mm) |
---|---|---|---|---|---|---|---|
DECT | |||||||
80 140 SN | 189 | 0.5 | 296 | 0.9 | DE body/angio Q30f | 0.5 | 32 × 0.6 |
SECT | |||||||
70 | 189 | 0.5 | 296 | 0.9 | DE body/angio Q30f | 0.5 | 32 × 0.6 |
80 | |||||||
100 | |||||||
120 | |||||||
140 |
The resulting images were analyzed using the Siemens syngo.via software (Version: VB50). A central region of interest (ROI) was adopted to avoid edge effects that might lower the CT number [45]. The virtual monochromatic CT number (in HU) and standard deviation (SD) were measured using the manufacturers software in 10-keV increments from 40 to 190 keV, using the same ROI. This was measured and averaged across the transaxial, coronal and sagittal planes for the cubic samples. The cuboid samples with rounded edges were measured in the coronal plane only. The electron density relative to water and measured in HU [46] (HURho), and the Zeff were measured on the Rho/Z application profile in syngo.via for all samples using ROIs in the DECT images. The ρe was then calculated using Equation (3) [47];
Measurement of the CT number and SD was performed for the SECT scans at each kV in the same way as DECT.
Line profiles were measured through the samples, in the scan direction, to check for uniformity of the print, using ROIs drawn in the center of the sample.
Results
Review of published data
When assessing radiological properties of 3D-printed materials suitable for CT imaging phantoms, the CT number is most commonly used [4], due to the relative ease in measuring the value for a given material. For the studies reviewed, the materials, printer, density of the material used and the CT values (HU) with the measured SD, where available from the papers studied, at different kV, are listed in Table 3.
Print material | Density | CT value (HU) | Ref. | |||
---|---|---|---|---|---|---|
P (g/cm3) | 80 kV | 100 kV | 120 kV | 140 kV | ||
Printer: 3D touch (3D Systems Inc. [48]) | ||||||
ABS (90% infill density) | -147 ± 18 | -113 ± 3 | [47] | |||
ABS (70% infill density) | -417 ± 17 | -408 ± 1 | [47] | |||
ABS (50% infill density) | -596 ± 22 | -580 ± 1 | [47] | |||
ABS (30% infill density) | -760 ± 13 | -743 ± 4 | [47] | |||
ABS (10% infill density) | -926 ± 14 | -901 ± 1 | [47] | |||
PLA (90% infill density) | -36 ± 7 | 8 ± 4 | [47] | |||
PLA (70% infill density) | -228 ± 1 | -234 ± 5 | [47] | |||
PLA (50% infill density) | -461 ± 3 | -466 ± 8 | [47] | |||
PLA (30% infill density) | -694 ± 10 | -688 ± 3 | [47] | |||
PLA (10% infill density) | -943 ± 14 | -916 ± 1 | [47] | |||
Zcorp 650 (3D Systems Inc. [48]) | ||||||
Gypsum (zp151 high-performance powder) | 1.57 | 372 to 995 (731) | [49] | |||
Printer: Lulzbot TAZ 5 (Lulzbot [50]) | ||||||
Photoluminescent PLA (90% infill density) | -36 ± 8 | -58 ± 1 | [47] | |||
Photoluminescent PLA (50% infill density) | -455 ± 22 | -460 ± 23 | [47] | |||
Photoluminescent PLA (30% infill density) | -673 ± 13 | -658 ± 14 | [47] | |||
Lulzbot hips-gy | -82.3 ± 13.5 | -61.4 ± 13.8 | -54.8 ± 13.7 | -36.6 ± 12.0 | [51] | |
e-SUN ABS-r | -85.6 ± 12.0 | -60.4 ± 12.7 | -49.6 ± 12.7 | -41.1 ± 10.0 | [51] | |
e-SUN ABS-bl | -84.3 ± 6.3 | -59.9 ± 5.2 | -45.2 ± 5.4 | -37.7 ± 5.1 | [51] | |
e-SUN ABS-w | -5.9 ± 7.6 | 2.0 ± 5.6 | 7.3 ± 5.0 | 10.5 ± 5.2 | [51] | |
Taulman nylon 680 | 24.6 ± 8.4 | 45.2 ± 6.7 | 59.1 ± 5.5 | 70.8 ± 5.6 | [51] | |
Ninjaflex-or | 55.2 ± 26.5 | 79.5 ± 24.5 | 100.4 ± 19.4 | 106.5 ± 8.5 | [51] | |
Polymer polyflex-w | 105.7 ± 27.0 | 122.7 ± 27.5 | 135.7 ± 27.4 | 142.3 ± 30.0 | [51] | |
Taulman t-glase-cl | 135.4 ± 16.1 | 153.6 ± 15.4 | 165.0 ± 16.3 | 173.1 ± 14.0 | [51] | |
e-SUN PLA-r | 151.2 ± 15.3 | 163.9 ± 14.6 | 168.5 ± 16.0 | 175.8 ± 13.1 | [51] | |
e-SUN PLA + -bk | 158.1 ± 24.3 | 165.9 ± 14.5 | 172.7 ± 39.8 | 177.4 ± 32.7 | [51] | |
Taulman t-glase-gr | 149.6 ± 9.7 | 166.6 ± 7.5 | 177.9 ± 6.5 | 188.5 ± 6.1 | [51] | |
Gizmodorks PLA-cl | 157.4 ± 9.4 | 178.0 ± 7.2 | 181.1 ± 6.2 | 192.4 ± 6.3 | [51] | |
e-SUN PVA | 211.0 ± 12.9 | 230.1 ± 15.0 | 229.1 ± 14.7 | 232.9 ± 12.9 | [51] | |
e-SUN PLA + -w | 345.0 ± 13.0 | 313.4 ± 8.7 | 298.7 ± 8.2 | 294.3 ± 8.5 | [51] | |
Printer: Replicator 2 (MakerBot Industries [52]) | ||||||
Woodfill (90% infill density) | 475 ± 14 | 474 ± 1 | [47] | |||
Copperfill (90% infill density) | 3575 ± 475 | 5372 ± 27 | [47] | |||
Bronzefill (90% infill density) | 3568 ± 532 | 7257 ± 24 | [47] | |||
Printer: Form 1 + (Formlabs [53]) | ||||||
Standard photopolymer resin | 40 ± 11 | 106 ± 1 | [47] | |||
Flexible photopolymer resin | 33 ± 10 | 98 ± 1 | [47] | |||
Printer: Object Connex (Stratasys Ltd. [54]) | ||||||
Acrylic polymer + TangoPlus AND VeroWhite | 1.29–1.39 | 97.8 ± 3.3 | [5] | |||
Printer: Connex3 (Stratasys Ltd. [54]) | ||||||
Transparent | 1.18 | 127 ± 15 | [14] | |||
Opaque (white) | 1.18 | 127 ± 15 | [14] | |||
Printer: Dimension Elite printer (Stratasys Ltd. [54]) | ||||||
ABSplus | 0.97 | -54 ± 13 | [21] | |||
Printer: Eden350 printer (Stratasys Ltd. [54]) | ||||||
FullCure720 | 104.8 | [22] | ||||
Tango Black FullCure920 | 72.2 | [22] | ||||
Printer: The Fortus 400 mc (Stratasys Ltd. [54] | ||||||
PC-ISO (polycarbonate) | -10 | [3] | ||||
Printer: Prusa i3 (Prusa Research [55]) | ||||||
Polylactic acid | 1.29 | [56] | ||||
Acrylonitrile butadiene styrene | 1.09 | [56] | ||||
Polyethylene terephthalate glycol | 1.30 | [56] | ||||
Thermoplastic Elastomers | 0.99 | [56] | ||||
Polyamide | 1.18 | [56] | ||||
Printer: Ultimaker 3 (Ultimaker [57] | ||||||
PLA (100% infill) | 1.23 | 183 ± 8 | [58] | |||
PLA (90% infill) | 1.04 | -21 ± 19 | [58] | |||
PLA (700% infill) | 0.83 | -228 ± 75 | [58] | |||
PLA (50% infill) | 0.62 | -446 ± 187 | [58] | |||
PLA (40% infill) | 0.48 | -545 ± 425 | [58] | |||
PLA (30% infill) | 0.38 | -661 ± 256 | [58] | |||
PLA-Fe (100% infill) | 1.87 | 2449 ± 45 | [58] | |||
PLA-Fe (90% infill) | 1.69 | 2153 ± 50 | [58] | |||
PLA-Fe (700% infill) | 1.40 | 1558 ± 152 | [58] | |||
PLA-Fe (50% infill) | 0.98 | 901 ± 421 | [58] | |||
PLA-Fe (30% infill) | 0.63 | 252 ± 675 | [58] | |||
Printer: Titan 1 3D printer (Kudo3D [59]) | ||||||
Acrylic polymer | 1.29–1.39 | 152.1 ± 3.5 | [5] | |||
SLS printer – model not specified (EOS [60]) | ||||||
Nylon (PA2200 (Polyamide 23) | 0.93 | -779 to -229 (-512) | [49] |
O'Reilly et al. [16] summarized tissue expected CT numbers over a range of kV – 80, 100, 120 and 140 kV – with SD, density, where known, and datasheets. The materials were scanned in air with a range of different shapes and sizes. The full data are available online for download and are therefore not included in this list.
Alqahtani et al. [24] published the properties of some tissues with their CT number and mass attenuation coefficient; however, the kV at which the measurements were made is unknown, and therefore the results have not been included. It was unclear as to the materials Mooney et al. [50] used for their experiments as the materials were named after the printer or the material name was stated without identifying the printer used, and again, the results were excluded.
Assessment of materials with DECT & SECT
The virtual monochromatic CT number obtained from DECT and the CT number measured for SECT, both at all measured energies, was plotted for the different materials against the measured relative ρe. For the materials with Zeff between 6.43 and 7.01, a linear correlation between the CT number and ρe was observed. The R2 value of the fit was 0.99 at monochromatic energies of 80 and 120 keV and of 0.90 at 120 kV (see Figure 1). At lower energies for these materials, there will be a combination of Compton and Photoelectric interactions, with Photoelectric interactions dependent on the cube of the atomic number [52]. At higher energies, Compton interactions, which are proportional to electron density [52], will be more dominant. The CT number measured at 120 kV will include lower energy contributions, and the differences in Zeff across the materials will be more apparent than at monochromatic energy of 120 keV, reducing the linear dependency. This is reflected in the R2 value, which increased with energy as the dependence on Zeff was reduced. In addition [52], there will also be more noise contribution at lower energies, which would also increase the uncertainty in the measurements [53].
The virtual monochromatic CT number is plotted against the range of keV for the cubic samples in Figure 2 and for the preprinted materials in Figure 3. The CT number increases with energy for all samples, with the exception of the Rigid 400 Resin sample, which decreases with increased keV. For all datasets, the SD of the measured CT number was higher at lower keV.
A summary of measured values of Zeff and ρe for each material are in Table 4. As expected, materials with a higher ρe have higher corresponding CT numbers (Figures 2, 3 & 4). With the exception of the Rigid 4000 Resin sample, the Zeff of all samples was below that of water, and above that of Adipose, which may explain why the CT number increases with energy. The ρe of all samples is all higher than that of water. This is reflected in the measured CT numbers and shown in Figure 4. The Rigid 4000 Resin sample has Zeff and ρe higher than water, reflected in the higher CT numbers and the drop in CT number with keV.
Material | Effective atomic number (Zeff) | Electron density (ρe) relative to water | DECT CT number at 120 keV | SECT CT number at 120 kV (bold range quoted in literature) |
---|---|---|---|---|
Air | 8.0 | 0.0012 (0.004) | -1046 to -986 | |
Water | 7.42 | 1 (3.343) | -7 to 7 | |
Liver | 7.405† | 1.051† | 50 to 70 | |
Adipose | 6.302† | 0.949† | -60 to -150 | |
Muscle | 7.414† | 1.042† | 40 to 60 | |
Myocardium | 39 ± 19 | |||
Flexible V2 | 6.45 | 1.119 | 110 ± 9.8 | 77 ± 4.0 |
Rigid 4000 Resin | 8.45 | 1.275 | 292 ± 9.0 | 337 ± 5.0 |
Clear Resin | 6.43 | 1.162 | 152 ± 10.3 | 118 ± 6.2 |
Aorta | 6.96 | 1.116 | 114 ± 4.9 | 81 ± 10.2 |
Carotid | 6.91 | 1.103 | 97 ± 7.8 | 70 ± 9.3 |
Coronary | 7.01 | 1.099 | 102 ± 5.4 | 68 ± 9.4 |
Myocardium 5 | 6.61 | 1.033 | 27 ± 5.2 | -2 ± 4.8 |
Myocardium 4 | 6.65 | 1.030 | 27 ± 5.7 | -6 ± 6.3 |
Myocardium 3 | 6.62 | 1.018 | 14 ± 6.0 | -20 ± 10.9 |
Myocardium 2 | 6.65 | 1.011 | 7 ± 5.2 | -24 ± 4.7 |
Myocardium 1 | 6.59 | 1.002 | -3 ± 9.4 | -37 ± 4.9 |
Endur RGD450 | 6.7 | 1.140 | 139 ± 10.4 | 75 ± 8.1 |
VeroGray RGD850 | 6.81 | 1.131 | 133 ± 7.2 | 56 ± 8.7 |
VeroWhitePlus RGD835 | 6.84 | 1.130 | 127 ± 8.5 | 63 ± 7.6 |
VeroCyan RGD841 | 6.64 | 1.129 | 130 ± 10.6 | 63 ± 11.2 |
VeroYellow RGD836 | 6.67 | 1.128 | 124 ± 11.8 | 67 ± 11.1 |
VeroBlackPlus RGD875 | 6.61 | 1.128 | 125 ± 10.5 | 55 ± 10.4 |
RGD720 | 6.55 | 1.127 | 131 ± 9.8 | 66 ± 10.3 |
VeroClear RGD810 | 6.52 | 1.127 | 124 ± 8.6 | 58 ± 8.8 |
VeroMagenta RGD851 | 6.77 | 1.125 | 126 ± 8.7 | 63 ± 5.9 |
High Temp RGD525 | 6.72 | 1.122 | 124 ± 10.6 | 64 ± 9.8 |
Digital ABS Ivory RGD 515 & RGD531 | 6.72 | 1.120 | 112 ± 9.7 | 52 ± 10.2 |
VeroBlue RGD840 | 6.52 | 1.116 | 116 ± 11.9 | 56 ± 7.1 |
Digital ABS green RGD515 & RGD535 | 6.76 | 1.101 | 118 ± 7.2 | 49 ± 11.3 |
TangoGray FLX950 | 6.69 | 1.109 | 104 ± 13.0 | 49 ± 8.4 |
TangoBlack FLX973 | 6.59 | 1.099 | 88 ± 9.7 | 40 ± 9.6 |
DurusWhite RGD430 | 6.43 | 1.098 | 78 ± 4.3 | 43 ± 9.4 |
TangoBlackPlus FLX980 | 6.72 | 1.083 | 66 ± 6.3 | 10 ± 9.8 |
TangoPlus FLX930 | 6.75 | 1.075 | 61 ± 6.1 | 13 ± 7.9 |
The virtual monochromatic CT numbers measured using DECT at 120 and 80 keV are shown in Figure 4, alongside the values for SECT at 120 kV. The myocardium samples 1, 2 and 3 had CT numbers below that of water, and all myocardium samples were outside of the muscle and myocardium ranges (shown in green and violet in Figure 4) at SECT 120 kV. The majority of the samples for the Stratasys Connex3 Object 350 printer were within soft tissue ranges at SECT 120 kV.
Line profiles for four samples are shown in Supplementary Figure 2, measured across the scan direction. No printing defects or artifact was observed and the density of the material appeared consistent throughout.
Discussion
3D-printed technology can be used to develop custom imaging phantoms with relative ease, be it from patient scans or custom-built designs. One of the biggest challenges lies in selecting the most appropriate material for the job. Analysis of materials has been performed by many centers, with little standardization between them, as noted by Goodall et al. [59], with factors such as manufacturer, software and printing methodology all contributing to variations between centers. For example, inhomogeneity in printed samples due to printer technology may cause variations in the density and consequently, the CT number [60]. Standard printing processes are therefore required for medical phantoms. Standardization is also required for imaging to allow easy comparison and selection of materials, although these should still be validated locally because the CT number may differ among vendors [30].
Nevertheless, comparison of 3D-printable material, printed and scanned with different technologies, showed surprisingly similar results, as summarized in Mille et al. [49], who investigated PLA in printed blocks and CT scanned in air to mimic soft tissue, whereas Dancewicz [51] measured PLA in scatter material (within a phantom), both at different infill densities and using different printers. The measured CT number at the different infill density correlated well for both studies, as shown in Table 3, with a maximum of 29 HU difference at 90% infill. Bibb et al. [60] measured samples in air and tissue equivalent material. Comparison of the CT numbers showed little difference in the -360 to 320 HU range, the tissue equivalent range, with over 100 HU difference for CT numbers ~850 and 1150.
The majority of 3D-printed materials measured over a range of energies showed an increase in CT number with increased kV. The Zeff of these materials was all measured to be below that of water. The CT numbers measured for soft tissue and muscle decrease with increased kV, adipose tissue being the exception [30]. Adipose tissue has Zeff below that of water, with soft tissue and muscle all having Zeff above water [34,54]. The one 3D-printed material that had Zeff above water was also the only material where CT number decreased with kV.
Novel Myocardial and vascular materials (Stratasys), printed on the J750 Digital anatomy 3D printer, and samples for the Stratasys Connex3 Object 350 printer and Form 2 printer were imaged using DECT to assess their imaging properties for tissue equivalence.
For the materials analyzed in this study, at a monochromatic energy of 120 keV, there will be little photoelectric contribution and Compton scatter will be dominant, where the dependence is on ρe. At lower energies, there will be a more dominant photoelectric contribution, which is dependent on Zeff. This is highlighted in Figure 1 where the line of fit is poorer for SECT 120 kV.
The virtual monochromatic CT numbers are not directly comparable to the CT numbers measured in SECT because the energy recorded for SECT is the maximum energy and comprises a spectrum of which the effective energy is approximately 50 to 60% of the peak [52].
Figures 3 & 4 show increased variation in the CT numbers for energies below DECT 60 keV. To characterize the imaging response of a material accurately, it is necessary to characterize over the whole spectrum to ensure the results are applicable to different scanning protocols and units. This can be easily achieved by ensuring the Zeff and ρe are a match to the material being replicated.
The data sheet published by O'Reilly et al. [16] included data for some Stratasys samples measured between 80 and 140 kV, as well as a range of other 3D-printed materials. The CT number also increased with increased kV for the majority of materials tested, including all the Stratasys samples.
Local testing of any 3D-printed phantom is recommended to ensure material consistency [21]. However, published literature suggests that materials printed with different technologies and imaged with different scanners and methodologies showed surprisingly consistent results provided they are imaged at the same energy [8,12].
Although there are limitations to the measurement of Zeff and ρe using DECT for high atomic number materials – for example, when using contrast agents or images with metal artifacts – the materials scanned all had a measured Zeff of under 8.5. As previously discussed, several studies have shown the accuracy of the Zeff and ρe measurement using DECT [34–37]. Many studies have carried out calculations of Zeff, or measured the mass density of materials using Archimedes' principle [6]. Measurement of these parameters is time-consuming, whereas DECT readily provides all of the information required to match materials using more fundamental variables to those of interest clinically.
The use of ρe and Zeff should provide a more fundamental method of matching materials and ensure that attenuation properties, and thus CT numbers, match across a wide range of spectra. This also lends to accurate modeling of materials when using Monte Carlo techniques to assess phantom design prior to printing.
Limitations
The accuracy of the DECT in measuring Zeff and ρe has been assessed previously with positive results [34–37]; however, accuracy of this measurement could potentially be improved by centering each sample within the field of view [27]. Hua et al. [34] found that accuracy started to drop when samples were placed over 60 mm away from the isocenter and recommended imaging within the field of view. The samples in this study were scanned along the isocenter.
The novel myocardial and vascular materials were a cubic centimeter in volume because larger samples were not available. At this size, there may be inaccuracies in the measurement of the CT numbers. This was mitigated as far as possible by obtaining ROIs in the center of the samples. Profiles of the CT number through the samples appear to stabilize in the center (see Supplementary Figure 2).
Conclusion
3D-printed technology can be used to develop imaging phantoms with relative ease based on patient scans or custom-built designs. Applications of such phantoms include commissioning, quality assessment over time, comparison and compatibility between centers, scanners, vendors and software, optimization of images in terms of acquisition and reconstruction parameters. The phantoms can be tailored to suit the need – for example, in use with dynamic CT imaging where phantoms can be designed to assess the triggering of contrast, looking specifically at variables such as flow rate, contrast agent used and the contrast delivery rate and how the resultant clinical imaging can be affected by these variables. However, one of the biggest challenges in 3D-printing phantoms lies in selecting the most appropriate material. Analysis of materials has been performed by many centers, with little methodological standardization. CT numbers from the same material in different papers matched reasonably well, despite variations in imaging protocols and CT vendors. The use of DECT to measure the CT number provides information over a range of monochromatic energies but also allows for simple measurement of Zeff and ρe, which may be helpful in describing the radiological properties of the material, including matching to real tissue, not only for CT but also for nuclear medicine and radiotherapy. The virtual CT numbers obtained from DECT cannot be directly compared with CT numbers measured using SECT. If using DECT to assess Zeff and ρe of materials, the SE components of the CT should be reconstructed for measurement of CT numbers, for an accurate comparison to values found clinically. However, rapid characterization of the response of materials over a range of virtual monochromatic energies supports confidence in material selection before printing.
The majority of 3D-printable materials scanned in the literature and for this paper showed a response with increased CT energy similar to that of adipose tissue, rather than soft tissues and muscle. When choosing a material for phantom development, the response over the full energy range to be used should be considered.
Material should always be validated locally using available technology and imaging settings. Standardized imaging would offer more confidence when comparing values – for example, standardization of sample sizes, scanning in air or scatter material, standard imaging settings and processing of data. Development of standardized testing and reporting criteria would be beneficial given the potential growth in this area.
The samples printed on the J750 Digital anatomy 3D printer are designed to match the texture and feel of the myocardium; however, the CT numbers of these materials were lower than those associated clinically with muscle and the myocardium, having a Zeff below that of water, and are therefore not suitable mimics for CT imaging.
The majority of 3D-printed materials in the literature behave like adipose tissue over a range of energies in computed tomography (CT) imaging.
If 3D-printed imaging phantoms are used over a range of energies, measurement of the CT number at a single energy is insufficient to describe the behavior of the material accurately.
Dual energy CT provides a quick and easy assessment of the Zeff and ρe of a material to categorize its radiological properties.
Supplementary data
To view the supplementary data that accompany this paper please visit the journal website at: www.futuremedicine.com/doi/suppl/10.2217/3dp-2022-0019
Author contributions
L Rowley – guarantor of integrity of the study, literature research, experimental studies/data analysis and manuscript preparation; L Rowley, E Davies – study concepts and design. L Rowley, E Davies and E Chung – manuscript editing.
Acknowledgments
The authors thank Claire Robinson, Kelly Parke and Odette Thraves for assistance in computed tomography image acquisition and processing, and Caroline Thorpe for their assistance in image transfer.
Financial & competing interests disclosure
Four 3D-printed samples (cubic; 1 cm × 1 cm × 1 cm) of the coronary artery, carotid artery, aorta and five myocardium materials (myocardium 1, 2, 3, 4 and 5) were printed using the Stratasys J750 Digital Anatomy 3D Printer (n = 8 sets) provided to the author for this research by Aiju (Instituto Technológico de Producto Infantil y Ocio, Alicante, Spain).
No writing assistance was utilized in the production of this manuscript.
Open access
This work is licensed under the Attribution-NonCommercial-NoDerivatives 4.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-nd/4.0/
Papers of special note have been highlighted as: • of interest; •• of considerable interest
References
- 1. . 3D Printing in Medicine. Rybicki FJGrant GT (Eds). Springer International Publishing, Cham, Switzerland, 5–22 (2017). • Good resource for summarizing technology and uses of 3D printing in medicine.
- 2. . An assembled prototype multimaterial three-dimensional-printed model of the neck for computed tomography- and ultrasound-guided interventional procedures. J. Comput. Assist. Tomogr. 41(6), 941–948 (2017).
- 3. Evaluation of PC-ISO for customized, 3D printed, gynecologic 192Ir HDR brachytherapy applicators. J. Appl. Clin. Med. Phys. 16(1), 246–253 (2015).
- 4. . Recent advances on the development of phantoms using 3D printing for imaging with CT, MRI, PET, SPECT, and ultrasound. Med. Phys. 45(9), e740–e760 (2018). •• A review paper summarizing 3D-printing phantom development and materials.
- 5. et al. Characterization of 3D printing techniques: toward patient specific quality assurance spine-shaped phantom for stereotactic body radiation therapy. PLOS One 12(5), 1–12 (2017).
- 6. . Tissue-equivalent materials for construction of tomographic dosimetry phantoms in pediatric radiology. Med. Phys. 30(8), 2072–2081 (2003).
- 7. . Technical note: accurate replication of soft and bone tissues with 3D printing. Med. Phys. 47(5), 2206–2211 (2020).
- 8. . Fabrication of a pediatric torso phantom with multiple tissues represented using a dual nozzle thermoplastic 3D printer. J. Appl. Clin. Med. Phys. 21(11), 226–236 (2020).
- 9. Suitability of low density materials for 3D printing of physical breast phantoms. Phys. Med. Biol. 63(17), 175020 (2018).
- 10. ICRU Report 44: tissue substitutes in radiation dosimetry and measurement. J. ICRU 23(1), iii–v (1989).
- 11. Imaging properties of additive manufactured (3d printed) materials for potential use for phantom models. J. Digit Imaging 33(2), 456–464 (2020).
- 12. Radiological properties of 3D printed materials in kilovoltage and megavoltage photon beams. Physica Medica 38, 111–118 (2017). • Assessment of 3D-printed materials using computed tomography.
- 13. Using 3D printing techniques to create an anthropomorphic thorax phantom for medical imaging purposes. Med. Phys. 45(1), 92–100 (2018).
- 14. Abdo-Man: a 3D-printed anthropomorphic phantom for validating quantitative SIRT. EJNMMI Phys. 3(1), 17 (2016).
- 15. Three-dimensional physical modeling: applications and experience at Mayo Clinic. RadioGraphics 35(7), 1989–2006 (2015). • Good review paper on modeling tissues and organs with 3D printing.
- 16. Simulating tissues with 3D-printed and castable materials. J. Digit Imaging 33(5), 1280–1291 (2020).
- 17. . Evaluation of three-dimensional printed materials for simulation by computed tomography and ultrasound imaging. Simul. Healthc. 12(3), 182–188 (2017).
- 18. . Imaging properties of 3D printed materials: multi-energy CT of filament polymers. J. Digit Imaging 30(5), 572–575 (2017).
- 19. Soft 3D-printed phantom of the human kidney with collecting system. Ann. Biomed. Eng. 45(4), 963–972 (2017).
- 20. Evaluation of 3D printing materials for fabrication of a novel multi-functional 3D thyroid phantom for medical dosimetry and image quality. Radiat. Phys. Chem. 135(January), 106–112 (2017).
- 21. Organ-specific SPECT activity calibration using 3D printed phantoms for molecular radiotherapy dosimetry. EJNMMI Phys. 3(1), 12 (2016).
- 22. Stenosis quantification of coronary arteries in coronary vessel phantoms with second-generation dual-source CT: influence of measurement parameters and limitations. Am. J. Roentgenol. 201(2), W227–W234 (2013).
- 23. Development of matched virtual and physical breast phantoms based on patient data. Med. Phys. 42(7), 866805 (2013).
- 24. Design and implementation of a prototype head and neck phantom for the performance evaluation of gamma imaging systems. EJNMMI Phys. 4(19), 19 (2017).
- 25. . Recent advances on the development of phantoms using 3D printing for imaging with CT, MRI, PET, SPECT, and ultrasound. Med. Phys. 45(9), e740–e760 (2018). •• Good review paper for 3D-printed phantom development.
- 26. Total mass attenuation coefficient evaluation of ten materials commonly used to simulate human tissue. J. Phys. Conf. Ser. 249, 012029 (2010).
- 27. . Accuracy and reproducibility of effective atomic number and electron density measurements from sequential dual energy CT. Med. Phys. 48(7), 3525–3539 (2021).
- 28. . Review and history of photon cross section calculations. Phys. Med. Biol. 51(13), R245–R262 (2006).
- 29. . ICRP Publication 89. Basic anatomical and physiological data for use in radiological protection: reference values. Ann. ICRP 32(3-4), 5–265 (2002).
- 30. . Chapter 1: Introduction to CT. In: The CT Handbook: Optimising Protocols for Today's Feature-Rich Scanners. Medical Physics Publishing, WI USA (2020).
- 31. XCOM: Photon Cross Sections Database. National Institute of Standards and Technology. NBSIR 87-3597 (2010). www.nist.gov/pml/xcom-photon-cross-sections-database (Accessed 9 November 2021).
- 32. . The Phantoms of Medical and Health Physics. Med. Phys. 43(9), 5264 (2014).
- 33. . Chapter 1 – Structure of Matter. In: Handbook of Radiotherapy Physics. Mayles PNahum ARosenwald JC (Eds). CRC Press, FL USA (2007).
- 34. . Accuracy of electron density, effective atomic number, and iodine concentration determination with a dual-layer dual-energy computed tomography system. Med. Phys. 45(6), 2486–2497 (2018).
- 35. Dual-energy CT quantitative imaging: a comparison study between twin-beam and dual-source CT scanners. Med. Phys. 44(1), 171–179 (2017).
- 36. . Accuracies of the synthesized monochromatic CT numbers and effective atomic numbers obtained with a rapid kVp switching dual energy CT scanner. Med. Phys. 38(4), 2222–2232 (2011).
- 37. Synthesized effective atomic numbers for commercially available dual-energy CT. Rep. Pract. Oncol. Radiother. 25(4), 692–697 (2020).
- 38. Stratasys Ltd. J750 Digital Anatomy Printer – datasheet. 1–8 (2019).
- 39. Stratasys Ltd. Connex3 Objet500 and Objet350 Multi-Material 3D Printers. www.stratasys.com/3d-printers/objet-350-500-connex3 (Accessed 9 November 2021).
- 40. Formlabs: Form 2. https://formlabs.com/uk/3d-printers/form-2/ (Accessed 28 June 2022).
- 41. Aiju Instituto Tecnológico de Producto Infantil y Ocio | AIJU. www.aiju.es/en/ (Accessed 4 January2022).
- 42. Stratasys. Biomechanical Tests confirm the potential for 3D printing synthetic myocardium with the Stratasys J750 Digital Anatomy Printer. https://www.stratasys.com/contentassets/d5e0b8c0b3064b09b246178756beec43/wp_pj_medtronic-dap-myocardium_0220a.pdf?v=48f8a5 (Accessed January 2023).
- 43. Statasys and Jacobs Institute. Creating biomechanically accurate vascular models (2020). www.stratasys.com/en/resources/whitepapers/creating--biomechanically-accurate--vascular-models/
- 44. Stratasys Ltd. Digital materials data sheet (2013). www.stratasys.com/siteassets/materials/materials-catalog/polyjet-materials/digital-abs-plus/mss_pj_digitalmaterialsdatasheet_0617a.pdf
- 45. . Computed tomography characterisation of additive manufacturing materials. Med. Eng. Phys. 33(5), 590–596 (2011).
- 46. SIEMENS Healthineers, syngo.via: Instructions for Use – syngo. CT Dual Energy for Dual Source Dual Energy data. VB50 (2020).
- 47. . A simple formulation for deriving effective atomic numbers via electron density calibration from dual-energy CT data in the human body. Med. Phys. 44(6), 2293–2303 (2017).
- 48. Using 3D printing techniques to create an anthropomorphic thorax phantom for medical imaging purposes. Med. Phys. 45(1), 92–100 (2018).
- 49. . Fabrication of a pediatric torso phantom with multiple tissues represented using a dual nozzle thermoplastic 3D printer. J. Appl. Clin. Med. Phys. 21(11), 226–236 (2020).
- 50. . Evaluation of three-dimensional printed materials for simulation by computed tomography and ultrasound imaging. Simul. Healthc. 12(3), 182–188 (2017).
- 51. et al. Radiological properties of 3D printed materials in kilovoltage and megavoltage photon beams. Phys. Med. 38, 111–118 (2017). •• Assessment of 3D-printed materials with supplementary data for a wide range of materials.
- 52. . Chapter 1 – Radiation Physics. In: Farr's Physics for Medical Imaging (2nd Edition). Elsevier, PA, USA (2008).
- 53. Principles and applications of multienergy CT: Report of AAPM Task Group 291. Med. Phys. 47(7), e881–e912 (2020).
- 54. The Phantom Laboratory. Catphan® 500 and 600 Product Guide (2021). Available: https://static1.squarespace.com/static/5367b059e4b05a1adcd295c2/t/615ef40255dbd2709cd9cfbd/1633612803610/CTP500600ProductGuide20211006.pdf (Accessed4 January 2022).
- 55. . Computed Tomography: Fundamentals, System Technology, Image Quality, Applications (3rd Edition). Publicis Publishing, Erlangen, Germany (2011).
- 56. . Imaging properties of 3D printed materials: multi-energy CT of filament polymers. J. Digit. Imaging 30(5), 572–575 (2017).
- 57. . CT Attenuation values of blood and myocardium: rationale for accurate coronary artery calcifications detection with multi-detector CT. PLOS One 10(4), e0124175 (2015).
- 58. Evaluation of 3D printing materials for fabrication of a novel multi-functional 3D thyroid phantom for medical dosimetry and image quality. Radiat. Phys. Chem. 135, 106–112 (2017).
- 59. Investigation of the effects of spinal surgical implants on radiotherapy dosimetry: a study of 3D printed phantoms. Med. Phys. 48(8), 4586–4597 (2021).
- 60. . Computed tomography characterisation of additive manufacturing materials. Med. Eng. Phys. 33(5), 590–596 (2011).