If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Biomedical Physics Division, Faculty of Physics, University of Warsaw, 5 L. Pasteur Street, 02-093 Warsaw, PolandMedical Physics Department, The Maria Skłodowska-Curie National Research Institute of Oncology in Warsaw, 5 WK Roentgen Street, 02-781 Warsaw, Poland
Warsaw University of Technology, Faculty of Chemical and Process Engineering, Department of Biotechnology and Bioprocess Engineering, Waryńskiego 1, 00-645 Warsaw, PolandCentre of Advanced Materials and Technologies CEZAMAT, Poleczki 19, 02-822 Warsaw, Poland
Warsaw University of Technology, Faculty of Chemical and Process Engineering, Department of Biotechnology and Bioprocess Engineering, Waryńskiego 1, 00-645 Warsaw, Poland
TPU and PLA honeycomb structure with 5%, 10%, 15% and 20% infill was investigated.
•
Radiological and dosimetric characteristics of TPU and PLA plates are similar.
•
Differences in PDD for different infills were observed only in the build-up region.
•
Differences between measurements and calculations are greater for PLA than TPU.
Abstract
Purpose
Dosimetric characteristics of 3D-printed plates using different infill percentage and materials was the purpose of our study.
Methods
Test plates with 5%, 10%, 15% and 20% honeycomb structure infill were fabricated using TPU and PLA polymers. The Hounsfield unit distribution was determined using a Python script. Percentage Depth Dose (PDD) distribution in the build-up region was measured with the Markus plane-parallel ionization chamber for an open 10x10 cm2 field of 6 MV. PDD was measured at a depth of 1 mm, 5 mm, 10 mm and 15 mm. Measurements were compared with Eclipse treatment planning system calculations using AAA and Acuros XB algorithms.
Results
The mean HU for CT scans of 3D-printed TPU plates increased with percentage infill increase from −739 HU for 5% to −399 HU for 20%. Differences between the average HU for TPU and PLA did not exceed 2% for all percentage infills. Even using a plate with the lowest infill PDD at 1 mm depth increase from 44.7% (without a plate) to 76.9% for TPU and 76.6% for PLA. Infill percentage did not affect the dose at depths greater than 5 mm. Differences between measurements and TPS calculations were less than 4.1% for both materials, regardless of the infill percentage and depth.
Conclusions
The use of 3D-printed light boluses increases the dose in the build-up region, which was shown based on the dosimetric measurements and TPS calculations.
]. Currently, 3D printing technology has become increasingly popular because it allows creating anatomically matching elements and accessories using imaging data. It has been used in various application areas, such as dentistry [
]. Implementing 3D printing in cancer centers allows the creation of customised accessories necessary for therapy. It has been widely applied to several cases of brachytherapy, where printing applicators [
] demonstrate that high precision and reproducibility of catheter position can be achieved. Furthermore, 3D printed technology has been successfully used to create patient-specific accessories, improving electron therapy by increasing surface dose [
]. In particular, this technique allows printing boluses with variable shapes and complexity required to deliver the prescribed dose to superficial cancer lesions [
]. The build-up effect characterizing MV X-ray radiation provides a positive skin-sparing effect; however, the extra material called bolus is needed when delivering the prescribed dose to the superficial tumors is required. The disadvantage of the currently available commercial boluses is the difficulty in ensuring proper contact with an irregularly shaped patient's body. The lack of adhesion between the bolus and the patient's skin can significantly affect the dose distribution and reduce the dose coverage. This problem may be solved by implementing a low-density 3D printed bolus with a personalized shape.
For a 3D printed bolus to be safe and effective, each polymer used should be well characterised, especially in terms of its radiological and dosimetric properties. Implementing new material in the clinic requires comparing calculations in the treatment planning system (TPS) with measurements. Validation of polymer properties can also be determined using Monte Carlo methods [
]. However, most 3D printing materials have not been well characterized, which is crucial for medical applications. According to recent publications, PLA (polylactic acid) and ABS (acrylonitrile butadiene styrene) are popular and commonly used polymers in the field of radiotherapy [
] have expanded the investigation of 3D printing material properties to include copperfill, bronzefill, photoluminescent PLA, and woodfill. Despite the available research results, it is worth emphasising that the use of the same type of filament from a different manufacturer or different rolls of the same filament from the same manufacturer may affect the properties of printed objects. Biltekin et al. [
] showed that not only the type of polymer, but also the infill percentage, the pattern of infill, and the direction of printing can significantly change the radiological and dosimetric properties of the printout. Ricotti et al. [
] also reported that changing the physical structure of the printed object (infill) changed its density, resulting in different radiation interaction characteristics. Furthermore, the relationship between the 3D printed object and its physical properties also depends on the printer brand, the type of printer, and the printing technique. Pereira et al. [
] showed that HU distributions depended on printing parameters, such as extrusion temperature or filament feed rate.
The purpose of our study was to investigate low-mass 3D printed plates fabricated with different percentage infill, which in the future can be used to produce boluses. Test plates with 5%, 10%, 15% and 20% honeycomb structure infill were printed using thermoplastic polyurethane (TPU) and PLA polymers: both are nontoxic, easy to disinfect, and inexpensive materials. The Hounsfield unit (HU) distribution and physical properties were analysed for each plate. The influence of percentage infill on the dose distribution in the build-up region was studied. Furthermore, the measured and TPS calculated dose distributions were compared. The effect of inhomogeneous 3D printed plates infill on the dose distribution along the beam's central axis for the honeycomb structure was also investigated. To our knowledge, this is the first study investigating the radiological and dosimetric properties of a 3D printed object with different percentage infill using a TPU polymer, and then comparing it with well-known PLA.
2. Materials and methods
2.1 Studies on 3D-printed plates
2.1.1 Characteristics of 3D printed plates
For the fabrication of test plates, two polymers in the form of 3D printer extrusion polymers were used. We compared the ubiquitous 3D printing materials: 3.00 mm diameter PLA (ZMorph) and 1.75 mm diameter TPU (DevilDesign). All models have been printed with a ZMorph 2.0 S printer.
3D honeycomb structure plates (Supplementary material, Fig. S1) designed as 15 cm × 15 cm × 1 cm cuboids were printed with 5%, 10%, 15%, and 20% infill. They were printed in 195 °C/60 °C head/hotbed (PLA) and 245 °C/100 °C (TPU).
The printing process time, cost, and amount of material used for a given plate were noted. The infill patterns of 3D printed plates were studied based on computed tomography (CT) images and described by the empty space (inter filament) size within their structures.
2.1.2 Hounsfield unit estimation
Correct assignment of CT numbers is significant for accurate dose distribution calculations in TPS. The HU values of each plate and the distribution of the HU values inside the plates were measured based on CT. Each of the 3D printed plates was scanned twenty times at 120 kV with a Somatotom Sensation Open CT scanner (Siemens) available at our Institute. The scanning resolution was 1 mm × 1 mm with a 1.5 mm slice thickness. The experimental setup is described in detail in the next section. CT data was transferred to Eclipse TPS (version 15.6, Varian Medical System). The contours of the plates were created using the drawing tools available in the Eclipse contouring module. To calculate statistics and generate histograms of the HU distribution within a structure, an in-house Python script was prepared for complete HU analysis. The delineations created in TPS were extracted from CT images in DICOM format (Digital Imaging and Communications in Medicine) using the Pydicom module. The HU values of the voxels enclosed within those contours were exported as NumPy arrays using the scikit image module. Based on twenty sets of structures prepared for each plate (including wall thickness), the average values of mean, standard deviation, minimum, and maximum HU were calculated. Histograms of HU values were also obtained within each 3D-printed plate. Linear regression was performed to study the averaged values of the mean HU and its median as a function of the percentage infill.
In addition, to verify how the 3D printing material is interpreted in TPS, the electron density of the printed plates was read based on the conversion curve implemented in Eclipse TPS and compared with water. The CT calibration curve allows for an approximation of the plate's relative electron density (ReD) on the basis of CT numbers. Using the determined ReD values and knowing the value of water electron density (), electron density of each plate was calculated according to the formula:
(1)
where .
Density for each plate (8 plates in total) was estimated using the following equation:
(2)
where m is the mass of a plate and V is its volume, measured with a high-precision scale and a calliper, respectively.
2.2 Percentage depth dose (PDD) - measurements and calculations
2.2.1 Experimental setup
The experimental setup consisted of eleven 30 cm × 30 cm slabs made of RW3 material (PTW-Freiburg, Freiburg, Germany): eight slabs with 1 cm thicknesses, two slabs with 0.5 cm thicknesses and one slab with 1 mm thickness. Slabs with the thickness corresponding to the depth of the given measurement were placed above the chamber adaptation slab, which consisted of two slabs with thickness of 1 cm each. At least seven 1 cm thick slabs were placed below the chamber adaptation slab to provide full backscatter. 3D printed plates were placed on top. The CT scan of the experimental setup with 5% PLA with two slabs for the Markus chamber is presented in Fig. 1a.
Fig. 1CT transverse scan of the experimental setup with 5% PLA plate (a) and the variability of the honeycomb structure with 5% infill (b) in the transverse plane. In Fig. (c), cross sections with the biggest and smallest amount of polymer are marked with dashed and dotted lines, respectively.
Percentage depth dose measurements were performed for 6 MV photon beams generated with linear accelerator CL 2300C/D (Varian Medical System, Palo Alto, CA). Tissue-Phantom Ratio20,10 (TPR20,10) was 0.668. For the measurement of PDD, the Markus plane-parallel ionization chamber with a nominal sensitive volume of 0.055 cm3 (type 23343, PTW), connected to the Unidos (PTW) electrometer, was used. Each measurement using the ionization chamber was repeated three times, allowing us to estimate its uncertainty, resulting in a PDD accuracy of 0.01%. The angles of the gantry, collimator and table were set to 0°. The source surface distance (SSD) measured to the first slab was constant and equal to 90 cm. The plates were reordered to take measurements at different depths by inserting the Markus chamber adapter into the desired position. There was no need to change the table position. An open photon square field of 10 cm × 10 cm was used. The monitor units (MU) per measurement were 100 MU and corresponded to 1 Gy delivered to the calibration point (depth = 10 cm). The dose rate was set at 300 MU/min. The dose was measured on the central axis at a physical depth of 1 mm, 5 mm, 10 mm, and 15 mm for the experimental setup with and without plates. We decided not to measure a dose deeper than 15 mm because the depth dose characteristic in the build-up region was the only region of interest in our study. The Gerbi correction method was applied to the measured signal obtained from the Markus chamber to correct overdoses in the build-up region [
]. All percentage depth dose curves were normalized to the measured value obtained without a plate at a depth of 15 mm.
To assess the usefulness of 3D printed plates as boluses, the differences between PDD measured with and without plates for different infills were calculated using the following formula:
(3)
where:
- value of percentage depth dose measured at d depth with test plate,
- value of percentage depth dose measured at d depth without test plate.
2.2.3 Treatment planning system calculations
Based on CT data and the delineated set of structures for each experimental setup, treatment plans were prepared using Eclipse TPS (version 15.6, Varian Medical System) with the same conditions as during dosimetric measurements (beam geometry, SSD, MU). Dose calculations were performed using two algorithms: AAA (version 13.6.23) [
]. For Acuros XB, the dose to medium mode was used. The 1 mm calculation grid was used to compare TPS calculations with the measurements. A 2.5 mm grid, the same as used in clinical practice, was used to study how the inhomogeneities influence dose distribution calculations. Doses at a physical depth of 1 mm, 5 mm, 10 mm, and 15 mm on the beam's central axis were reported and compared with the PDD measured with the Marcus chamber. By analogy to measurements, all percentage depth dose curves were normalized to the dose calculated without a plate at 15 mm depth.
2.3 Influence of printing pattern inhomogeneities on dose distributions
The test plates were printed with a 3D honeycomb structure of 5%, 10%, 15% and 20%, resulting in inhomogeneous objects consisting of empty space and solid parts. The honeycomb shape proposed in this paper is a periodic structure that repeats with a specific frequency, depending on the percentage infill. For example, for the 5% infill plate, the repeatability of the structure on CT scans in the horizontal direction is every 1.5 cm. In Fig. 1b, the repeatability of the 5% infill is presented. The pattern at Z = 0.0 mm and Z = 15.0 mm is identical. The size of the empty space decreases with increasing infill percentage, resulting in a decrease in repeatability distance. The distance is equal to 0.75 cm, 0.51 cm, and 0.38 cm for the 10%, 15%, and 20% honeycomb structures, respectively. Because the beam may pass through random parts of the honeycomb structure, the maximum difference between the region with the largest and smallest amount of polymer was assessed. PDD calculations were performed for two central beam alignments as shown with dashed and dotted lines in Fig. 1c. for each slice (Z direction). Only slices within a single repetition of the honeycomb pattern were taken into account. The number of slices depended on the plate infill.
Differences between depth dose calculated under the maximum (Dpolymer) and minimum (Dempty) quantity of polymer were determined using the equation:
(4)
The values of were calculated for each testing plate at a physical depth of 1 mm, 5 mm, 10 mm, 15 mm and presented as box plots (Fig. 4).
3. Results
3.1 Characteristics of 3D printed plates
The time needed to print 3D TPU test plates was 2495 min, 3465 min, 4370 min, and 5185 min for 5%, 10%, 15% and 20%, respectively. As for TPU, a lower (20%) speed was used, and the time required for printing PLA plates was around 4–5 times shorter, depending on the percentage infill. As expected for both polymers, material consumption increases with increasing infill percentage (67.8 g–149.4 g for TPU plates). For the same infill, the polymer mass consumption for PLA was approximately 1.2–1.4 times higher than for TPU. The costs of TPU polymers were 3.19$, 4.25$, 5.43$, 6.51$ for 5%, 10%, 15% and 20%, respectively. The differences between the printing costs using TPU and PLA were estimated to be not greater than 1 $ for a given plate and were slightly higher in the case of TPU. The 3D printing technical properties of the TPU and PLA polymer test plates are described in Table S1, Supplementary material.
CT scans were performed for each 3D printed plate with 5%, 10%, 15%, and 20% infills (presented in the Supplementary material, Fig. S2) and used for the HU analysis. The average values of the mean, median, maximum, minimum and standard deviation of the HU obtained for twenty scans of each plate are presented in Table 1. Increasing the infill percentage corresponds to higher values of the minimum and maximum HU, both for TPU and PLA. The mean and median HU increases with an increasing infill percentage, regardless of the material used. The increase in plate infill percentage corresponds to the decrease in STD of the HU distribution for both TPU and PLA. However, the STD values for TPU are slightly higher.
Table 1Average descriptive statistics of the Hounsfield unit distribution for different types of 3D printed plate.
Histograms of HU values are presented in Fig. S3, in the Supplementary material. In the case of plates with 5% and 10% infill, a large peak of approximately −1000 HU was observed for both materials. The peak resulting from empty space inside the plate is lower in the case of plates with 10% infill compared to plates with 5% (the number of counts is 4.29 and 3.33 times smaller, respectively, for TPU and PLA). The HU distributions show the multimodality due to 3D printed plate structure and scanning artefacts. The FWHM of the HU distribution decreases with increasing infill percentage, taking the smallest values for 20% PLA plates.
Linear regression was used to estimate the relationships between the average value of mean HU and percentage infill (Supplementary material, Fig. S4). The linear equations were fitted using the least squares method: y = 22.6x − 849.3 for TPU and y = 22.9x − 859.7 for PLA. The goodness of fit is equal to R2 = 0.99994 and R2 = 0.9990 for TPU and PLA, respectively. The uncertainties of the fit parameters are below 3%. The average value of the mean HU increased linearly with an increase in the infill percentage, regardless of the material used.
Furthermore, averaged median values were calculated for PLA plates and used to fit the linear function y = ax + b. By performing a linear regression with R2 = 0.9560, the following parameters with uncertainties were obtained based on our data: a = 25.9 ± 3.9, b = -875 ± 55. These fitted parameters were compared with the results of Ricotti et al. [
] (a = 16.20 ± 0.79, b = -895 ± 30). According to the 3sigma test, the parameters are the same within the uncertainties.
The physical properties of the materials analyzed are listed in Table 2. While the electron density relative to water was estimated based on the HU calibration curve implemented in TPS, the electron density of 3D printed plates and their mass density were calculated according to Equations ((1), (2)). The values for water are shown as a reference. Using our clinical CT data, we found that an increase in the infill percentage corresponded to a higher electron density; however, for the same infill percentage, the electron densities for the TPU plates were lower. The mass density of the test plates increased with an increasing infill percentage for both materials. Furthermore, the mass density obtained for plates with 20%, the highest used in this study infill, was lower than that for water (0.72 for TPU and 0.68 for PLA).
Table 2Physical properties of 3D printed test plates and water.
The percentage depth dose curves measured in the build-up region for TPU and PLA plates with different infills are shown in Fig. 2. Differences in the PDD values for different infills of the testing plates were observed only at depths of 1 mm and 5 mm. The PDD increased with increasing infill for both bolus materials. The difference between PDD for 5% and 20% infill plates at 1 mm depth was 12% and 14% for TPU and PLA, respectively. At 5 mm depth, the PDD differences between maximum and minimum infill did not exceed 4.5% for both materials. At depth mm, the measured PDD values were very similar (99%−103%), regardless of the plate infill for both materials. The PDD values measured for PLA and TPU plates at a depth of 1 mm did not differ by more than 2%. For measurements performed at depths equal to 5 mm and deeper, the differences between the PDD values for PLA and TPU were less than 3%. Furthermore, Fig. 2 shows PDD measurements performed without a 3D printed plate to emphasize the influence of the plates on the PDD distribution in the build-up region.
Fig. 2Percentage depth dose measured in the build-up region for TPU (a) and PLA (b) test plates printed with different infill. The PDD measured without a plate was also presented.
A 5% infill plate showed a PDD increase of 32% at 1 mm compared with the absence of bolus. For 20% infill plate, the results increased by around 45%. The decreased with increasing depth for each testing plate, and at mm depth, they were less than 5%, regardless of the infill percentage and the type of material. All calculated values of the are included in the Supplementary material (Table S2). Regardless of the infill and material used, PDD increased from approximately 45% to 91% for 1 mm and from 85% to 102% at 5 mm.
Fig. 3 shows the differences given in percentage points between the measured and calculated PDD with and without test plates for two algorithms implemented in Eclipse TPS. At 1 mm depth, a better agreement between the measurements and the calculation was obtained for the Acuros XB algorithm than for the AAA algorithm. For depths greater than 1 mm, the differences for both algorithms were small and less than 3.90%. The most significant differences were observed between measurements and calculations for fields irradiated without a test plate at 1 mm depth. Differences of 4.70% and 7.10% were obtained for the PDD calculated by Acuros XB and AAA, respectively.
Fig. 3Differences between measured and calculated PDD obtained with and without 3D printed plate. Calculations were performed using two algorithms: AAA ((a) for TPU and (b) for PLA) and Acuros XB ((c) for TPU and (d) for PLA).
Fig. 4Differences between depth doses calculated along the central axis of the beam setting on the cross section with a maximum and minimum amount of the polymer. Calculations were performed using AAA ((a), (b) for TPU and PLA, respectively) and Acuros XB ((c), (d) for TPU and PLA, respectively) algorithms.
Fig. 4 shows the differences between PDD along the beam's central axis, passing through the maximum and minimum quantity of the polymer, calculated according to equation (4). These differences are presented as box plots showing the median, minimum, and maximum values and the corresponding quartiles. The distributions of values at different depths determined on CT scans of each 3D printed plate were presented for all plate infills, two materials and two Eclipse algorithms.
Generally, the differences calculated using Acuros oscillated around zero, whereas AAA calculations gave negative values, which means that the Dempty is smaller than the Dpolymer. For low infills (less than 15%), the differences calculated at depths < 10 mm were smaller for PLA, independently of the algorithms (excluding the calculations for AAA at 1 mm and 10% infill). The rest of the calculations gave differences of less than 2%.
4. Discussion
Easy access and low cost result in the growing popularity and broad application of 3D printers in medicine, including radiation therapy. One of the topics of interest is the replacement of conventional boluses with printed ones. 3D printed boluses have many advantages over commercial ones. Flexible bolus materials (e.g. Tango or Agilus) perfectly adhere to the skin, especially with complex and irregular shapes of the patient's body. In addition, by using a malleable polymer, patient discomfort associated with wearing the bolus can be reduced. The disadvantages of a material like Agilus are that it is expensive and is not easily accessible. The use of hard TPU and PLA polymer as a bolus can cause patient discomfort, it requires a long time for printing and the need for individual object characterization, but the easy access and low cost of these materials are a huge advantage in the context of clinical use. The best solution would be a hard bolus placed on the patient's skin providing a high surface dose, which is the goal of our future research. Thanks to 3D printed boluses, it is possible to print a complex, individually shaped bolus, ideally suited even to a very irregular patient's body surface. The problem with inter-fraction reproducibility and intra-fraction changes in the bolus shape and position can be minimised so that unwanted air gaps affect the surface dose less. Furthermore, bolus misplacement can be avoided during treatment preparation. Due to the difficulty in reproducing boluses planned in TPS, it is recommended to collect CT with a previously prepared bolus. The type, thickness, and size of the boluse are fitted to the patient’s body prior to CT scanning. Collecting CT data with the bolus allows any bolus imperfections to be taken into account when calculating the dose. Unfortunately, following this procedure, it happens that after the planning target volume (PTV) delineation, the bolus is too short or too thin. This makes it impossible to prepare an acceptable treatment plan and administer the prescribed dose to the entire PTV volume. The bolus can also be too large, leading to loss of the skin-sparing effect and delivery of an unnecessary dose to healthy tissue.
Although the benefits of using 3D-printed boluses are many, it is worth emphasizing that before a given material is used in clinical practice, its properties must be characterized, and CT imaging and treatment planning systems should be adapted to its application. This study aimed to evaluate the radiological and physical characteristics of a 3D printed test plate with a honeycomb pattern and different infill percentages. Two materials, TPU and PLA, were used. We have decided not to use over 20% volume infill because of a significant increase in the mass of models. A low bolus mass may be more beneficial in clinical use, e.g. for patients treated with the DIBH (Deep Inspiration Breath-Hold) technique. Furthermore, the influence of low-density plates on the dose distribution in the build-up region for X6 MV was investigated to evaluate clinical usefulness. Finally, the accuracy of the calculations in Eclipse TPS for two calculation algorithms (AAA, Acuros XB) was compared with measurements.
The time needed to print 15x15x1 cm3 cuboid TPU plates was around 4–5 times longer than PLA plates due to their high elasticity (the TPU polymer was sliding out of the toolhead). The printing time depends on the characteristics of the material and the printer. High printing speed for flexible TPU is not possible due to the toolhead construction of the Zmorph 2.0S. The HU histogram was flat and wide for the lower infills, while the shape of the histograms for the 20% infill was close to the Gaussian distribution around the mean HU.. The mean value of HU for 5% infill was around −739 and −731 for TPU and PLA, respectively, while the mean value of HU for 20% was −399 and −407 for TPU and PLA, respectively. No significant differences were observed between the mean HU determined for both materials. The increase in plate infill percentage corresponds to a decrease in STD of the HU distribution for both TPU and PLA, which means that a higher infill percentage implies higher density and homogeneity within the test plates. Linear regression equations for the mean HU and the infill percentage were determined for 1 cm thick plates printed with PLA and TPU in a honeycomb structure. Together with PDD measurements, these equations might be helpful in clinical use to select the appropriate infill to achieve the necessary dose in the build-up region. The choice of bolus infill is a compromise between time and printing cost that correlates with the percentage fill and dose coverage on the patient's skin, which is not always required to be 100%.
The mass densities of raw TPU and raw PLA are similar and equal to approximately 1.23 g/cm3, while the gel bolus used clinically in our hospital is characterized by 1.03 g/cm3. However, the average bolus mass density is gradually reduced when printed with a less than 100% infill. The low mass density boluses can be beneficial in the DIBH technique or postoperative radiotherapy when the surgical scar remains unhealed. Due to the low mass of the bolus, placing it on patients’ sensitive or painful skin reduces the pain associated with pressure on the wound. In addition, placing such a bolus on the chest of patients treated with the DIBH technique may cause less difficulty in maintaining a deep breath than with a heavier bolus.
PDDs for TPU and PLA were measured. The higher the infill percentage, the higher the dose at 1 mm and 5 mm depth. At higher depths measured PDDS were higher that 95%, regardless of the infill percentage and the material’s type. The PDD without plates was also measured and compared with the measurements with 3D printed plates. Increasing the depth of the measurements decreases the differences between the PDD measured with and without the plate.
Furthermore, the differences between measurements and calculations using the AAA and Acuros XB algorithms implemented in Eclipse TPS were calculated and compared for TPU and PLA materials. Since the calculation of the build-up region is unreliable and do not give an accurate value of the surface dose, they were used as a relative evaluation. Many groups have reported that the AAA algorithm implemented in Eclipse TPS cannot precisely calculate the surface and near-surface dose. The AAA algorithm tends to overestimate or underestimate the surface and near-surface dose [
], depending on the version of the algorithm (the older versions underestimate the dose, while the newer versions overestimate). Studies on the Acuros XB algorithm are mostly concerned with calculations in inhomogeneous media, which rarely refer to the dose in the build-up region. Published research shows that the Acuros XB algorithm is closer to Monte Carlo calculation and dosimetry measurements than the AAA algorithm in the context of surface and near-surface dose [
The Investigation of The Anisotropic Analytical Algorithm (AAA) and the Acuros XB (AXB) Dose Calculation Algorithms Accuracy in Surface and Buildup Region for 6 MV Photon Beam Using Gafchromic EBT3 Film.
]. Referring to the results of our study, the differences at 1 mm depth were less than 3.90% for TPU and 4.10% for PLA, while at >1 mm depth they were less than 3.90% for both materials. The influence of the TPS calculation algorithm and the size of the calculation grid on the obtained results cannot be neglected.
Based on the analysis of honeycomb infill inhomogeneity, we conclude that for 20% infill, a satisfactory therapeutic dose can be obtained close to the surface of the patient's skin. Additionally, dose discrepancies results from printing inhomogeneity are acceptable and less than 0.5% (except in 3 calculation points, where the dose discrepancy is less than 1.2%). Similar results were obtained for both algorithms regardless of the printing material and the calculation depth. The largest interquartile range was observed for the 5% infill due to the size of the empty space inside the infill pattern. In the case of not full bolus infill, some dose differences appear depending on a cross-section of the bolus, and they are more pronounced for lower infill percentages.
Many studies have been conducted on the application of 3D printing in radiotherapy and polymer properties. Based on the available literature, it can be concluded that many factors, such as the 3D printing technique, the type of printer, and the polymer used, can affect the HU, electron density, or the shift of PDD. The same type of polymer produced by different manufacturers may have different characteristics, resulting in different bolus properties [
B. Wittbrodt J. Pearce The effects of PLA color on material properties of 3-D printed components Addit Manuf. 8 2015 110 116 10.1016/j.addma.2015.09.006. comcomponentscomponents. Addit Manuf. 2015; 8:110-116. doi:10.1016/j.addma.2015.09.006.
]. Despite many published works on the application of 3D printing in radiation therapy, before implementing 3D printing for clinical use, numerous studies should be carried out using their own equipment and polymers.
5. Conclusions
To conclude the results of our study, there are no significant differences between the radiological and dosimetric characteristics of TPU and PLA 3D printed plates, except for the printing time. Moreover, bolus with the 20% infill printed with honeycomb pattern is the best compromise between price, weight, and ensuring a satisfactory surface dose. Despite the inhomogeneity of the honeycomb structure, dose discrepancies are negligible. The low cost of 3D printed plate preparation, relatively easy access to 3D printers, and the ensuring a satisfactory surface dose give the potential for the bolus to be used for patients treated with radiation therapy.
Funding
This research did not receive specific grants from funding agencies in the public, commercial or non-profit sectors.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
The authors gratefully acknowledge the contributions of Mariusz Gruda for help and support.
Appendix A. Supplementary data
The following are the Supplementary data to this article:
The Investigation of The Anisotropic Analytical Algorithm (AAA) and the Acuros XB (AXB) Dose Calculation Algorithms Accuracy in Surface and Buildup Region for 6 MV Photon Beam Using Gafchromic EBT3 Film.
B. Wittbrodt J. Pearce The effects of PLA color on material properties of 3-D printed components Addit Manuf. 8 2015 110 116 10.1016/j.addma.2015.09.006. comcomponentscomponents. Addit Manuf. 2015; 8:110-116. doi:10.1016/j.addma.2015.09.006.
00077-7&id=gr1.jpg){kind=link}
00077-7&id=gr2.jpg){kind=link}
00077-7&id=gr3.jpg){kind=link}
00077-7&id=gr4.jpg){kind=link}