© A. Chaikh et al., Published by EDP Sciences 2025

1 Introduction

While IRSN has extensive experience with fetal exposure during radiological or nuclear medicine examinations (Étard and Aubert, 2009), fetal radiation protection during radiotherapy of pregnant women remains rare and understudied. Fetal dose assessment for pregnant women undergoing radiotherapy is challenging due to several factors. Firstly, the fetus is located outside the treated volume, where treatment planning systems cannot accurately calculate the dose. Secondly, currently commercially available phantoms do not simulate pregnant women. For radiotherapy, healthcare professionals are faced with situations where they must find a balance between improving the patient’s survival and quality of life while ensuring radiation protection for the fetus. The French society for radiation oncology updated its recommendations on radiotherapy and pregnancy in 2022 (Michalet et al., 2022). They particularly recommend a phantom-based estimation of the dose delivered to the fetus and its optimization. To our knowledge, there is no commercially available phantom that allows the evaluation of fetal dose for radiotherapy. However, a phantom as close as possible to the anatomy of a pregnant patient is preferable. Indeed, the use of a female human phantom or an equivalent phantom, without any additional tissue material over the phantom’s abdomen to simulate pregnancy, may introduce unknown inaccuracies in the fetal dose estimation (Mazonakis and Damilakis, 2017). An accurate estimation of fetal dose is necessary to evaluate the risks of radiation exposure for the fetus (ICRP No.84, 2000) (Streffer et al., 2003) and enables the optimization of fetal dose by comparing, for example, the impact on this dose of different treatment parameters (i.e., beam configuration, energy, etc) or treatment modalities (Noël et al., 2006). According to the literature, only a few physical phantoms of pregnant women have been developed to evaluate fetal dose during radiological examinations or radiotherapy (Hoerner et al., 2018; Shirkhani et al., 2020 ; Matsunaga et al., 2021; Kopačin et al., 2023). In the current study, 3D printed “bellies” of pregnant woman has been developed to be added on an ATOM® woman phantom, mimicking different stages of pregnancy and enabling fetal dose estimation.

3D printing technology has been used to produce individualized phantoms, quality assurance phantoms, compensators, brachytherapy applicators as well as equipment supporting patient immobilization (Gear et al., 2014; Su et al., 2014; Arenas et al., 2017; Rooney et al., 2020). For these applications, materials such as thermoplastics or light-cured resins are commonly used. However, special consideration must be given to the accuracy of the printed materials, particularly regarding printing precision and tissue equivalence.

The present study aims to describe a method of 3D modeling and printing bellies which will be added to the ATOM® phantom. Additionally, the study aims to evaluate the dosimetric water equivalence of the VeroWhite resin for radiotherapy beam energies.

2 Materials and methods

2.1 ATOM® phantom

The anthropomorphic adult female ATOM® phantom from Computerized Imaging Reference Systems (CIRS) consists of three parts: the head with the cervical spine (plates 1-10), thorax (plates 11-23) and pelvis (plates 24-38)¹, the plates being made of materials reproducing the attenuation of human tissues. The compositions and nominal physical densities of the materials used in the phantom ranged from (0.9 to 1.1 g/cm³)². The height and the weight of the phantom were, respectively, 160 cm and 55 kg, leading to body mass index (BMI) of 21.5 kg/m².

2.2 3D-printing process to model the bellies for different stages of pregnancy

First, the ATOM® phantom was scanned to obtain its external contours (GE Discovery RT, 120 kV, X-ray tube current from 148 to 379 mA and slice thickness of 0.625 mm). Then, the DICOM images were imported into 3D Slicer software to extract the external contours in (0.3 dcm) format (Fedorov et al., 2012; Ashenafi et al., 2023). These were then imported into Rhinoceros® 3D software³. The Non-Uniform Rational Basics Spline (NURBS) format was chosen. The NURBS format was used for phantom modeling, because of their flexibility and accuracy for the various transformation operations. The bellies for different stages of pregnancy were modeled to fit perfectly on the ATOM® phantom and to respect the fundal height, as described in the next section. The bellies represent the volumes to be fixed on the pelvis section of the ATOM® phantom. Then, the NURBS surfaces were converted to mesh using the Stereo Lithographic file format (*.stl) which can be easily read by the 3D printer. Finally, the 3D modeled volumes were printed using 3D Printer (Stratasys), and VeroWhite resin (Polyjet Opaque Resin)⁴. This technique uses photosensitive polymer liquids to create 3D printed objects. The liquid undergoes photopolymerization meaning it is solidified by UV light, layer by layer, to create 3D prints.

2.3 Modeling, geometry of the bellies and definition of points of interest for fetal dose measurements

The symphysis fundal height is an important clinical indicator to assess fetal growth and development and is used primarily to detect fetal intrauterine growth restriction (Robert Peter et al., 2015). It represents the distance, in centimeters, from the mother’s pubic bone as a reference point to the top of the uterus (fundus). In practice, from the 4th to the 7th month, to obtain the reference measurement based on the pregnancy term, the number of months is multiplied by 4. From the 8th month until term, the number of months is multiplied by 4, and the resulting number is then subtracted by 2 cm (Ayet, 2014).

Three stages of pregnancy were considered for this study: first, 2^nd, and 3^rd trimesters. The previous formula was used to determine the fundal heights for the 2nd, 6th and 8-months of pregnancy, corresponding to the three stages. To define the 3D bellies’s dimensions, the theoretical fundal height, and the dimensions of the ATOM® phantom in x and y axis were used as fixed values, then the bellies dimensions were manually optimized in respect of these values and to be realistic. In collaboration with a gynecologist, the position of the uterus on a lateral section of the ATOM® phantom with bellies, was determined in the form of ellipses.

Different points inside these elliptic uterus were chosen for measurements : at the level of the fundus and cervix corresponding either to the fetus’s gonads or its head, one point was selected to measure the maximum dose due to scattered radiation from the patient (a point deep closest to the woman’s head and in the uterus) and additional points were defined in the uterus to assess the average fetal dose (Fig. 1).

Because a part of the uterus is located within the ATOM® phantom, two specific ATOM® phantom plates were replaced by printed plates to enable ionization chamber point measurements. This was achieved by creating holes in them.

Fig. 1 Representation of the 3D modeling bellies placed over the ATOM® phantom. The fundal heights value and its representation are shown in red, and the green points present the holes used to place the ionization chamber and measured dose.

2.4 Characterization of the VeroWhite resin

VeroWhite resin was initially motivated in this study by the following reasons: resin-based printing helps to minimize air gaps between layers unlike filament-based methods, does not require manual post-printing treatment, and its density is relatively close to that of water compared to other available resins. The VeroWhite resin has already been used for the design of thorax phantom with mobile tumor for radiation dosimetry (Mayer et al., 2015) and to assess the thyroid dose to children using ¹³¹I (Beaumont et al., 2017).

2.4.1 Chemical composition and physical density

The chemical composition of the resin was determined by a CHONS (Flashsmart model, Thermoscientific) analyzer. Few milligrams of resin sample is weighed with a microbalance into a tin capsule, which is then introduced into an oxidation/reduction reactor heated to 950°C. A controlled supply of oxygen causes the tin to oxidize, an exothermic reaction that raises the temperature to 1800°C. At this temperature, organic and inorganic substances are converted into gases (CO₂, N₂, H₂O, SO₂) which, after reduction, are separated on a gas chromatography column and finally detected by a thermal conductivity detector. The oxygen is quantified in a special pyrolysis reactor heated to 1065°C in an oxygen-free environment. A catalyst ensures the conversion of oxygen into CO, which is separated and detected with the gas chromatography with thermal conductivity detector.

The density of the resin was measured using the helium pycnometry method. Gas pycnometry (AccuPyc II 1340 model, MICROMERITICS) is used to determine the true – or skeletal – density of solid materials, which is the mass of solid material divided by its volume minus the free volume. A sample is placed in a chamber of known volume, which is sealed and pressurized. The gas fills the empty spaces within and between the sample particles. The sample chamber is then expanded to an adjoining reference chamber of known volume. The change in pressure is used to calculate the volume of the sample. Skeletal density is calculated from the sample mass and the volume it occupies. This method is useful for determining the skeletal density of materials, even those with small pores and irregular shapes.

2.4.2 Calculated and measured attenuation coefficients

To assess the water dosimetric equivalence of the VeroWhite resin, two distinct methods were used: by calculation means using the chemical composition and by measurements of the attenuation.

In the first method, the percentage of each chemical element is used to determine the linear attenuation coefficients (µ) as a function of photon energy. The linear attenuation coefficients of each chemical element were taken from data available on the NIST (National Institute of Standards and Technology) website⁵.

In the second method, the measurements were carried out using Synergy Elekta accelerator with 6 MV photon beam whom quality index is 0.685. In this case, a semiflex ionization chamber 31010 (0.125 cc) was positioned at a depth of 3 cm within a block of 5 cm RW3 plates, as described in Figure 2. Then, various thicknesses (up to 21 cm) of RW3 plates or VeroWhite printed plates were placed above the block to measure the attenuation of the beam. The measured charge(I₀) without added plates was taken as the reference one. Then, for each thickness (x) of RW3 or printed plates, the collected charge (I) was measured. The attenuation coefficient can be calculated, as the collected charge is directly proportional to the incident fluence, assuming that the beam spectrum is identical for the various measurements performed. The linear attenuation coefficient in cm⁻¹ (μ) was calculated as given by the Beer–Lambert’s law (Alshipli et al., 2018) using the following equation: $$\mu =\frac{1}{x}\times ln\left(\frac{I_0}{I}\right)$$(1)

However, it should be noted that the experimental conditions differ slightly between the two sets of measurements due to the difference in dimensions between the RW3 plates and the available VeroWhite resin elements (Fig. 2). The RW3 plates have a volume of 30 × 30 × 1 cm³ whereas the printed plates of VeroWhite have a volume of 3 × 3 × e cm³, where (e)represents the given thickness. The beam size at the isocenter was 3 × 3 cm² in order to meet the minimum required size according to the documentation of the ionization chamber used and to always be smaller than the dimensions of the crossed VeroWhite plates.

Fig. 2 Experimental setup to measure and compare linear attenuation coefficient using RW3 plates (left) and VeroWhite resin (right) with a 6 MV Elekta beam.

3 Results

3.1 Modeling bellies for different stages of pregnancy

The ATOM phantom alone represents the first trimester of pregnancy. For a 2-month pregnancy, the fundal height is 8 cm. The printed bellies simulate the 6th and 8th months of pregnancy with fundal heights of 24 and 30 cm, respectively. They were printed to be complementary, that means that the belly of the 3rd trimester is placed above the one of the second trimester. Furthermore, to meet the size constraints of the 3D printer, the belly corresponding to the last trimester is cut into two parts, as well as the printed plates.

For the first trimester, one plate was modeled at the uterus level (ATOM plate number 30) and printed using the dimension of the ATOM® phantom. For 2^nd, and 3^rd trimesters, two bellies and one other plate (ATOM plate number 25) were modeled and printed, as visualized in Figure 3.

The defined holes were modeled to insert the Farmer (PTW 30013) ionization chamber (0.6 cc). Note that, each hole has a rod to be plugged in case of non-use. To prevent any difficulty to insert the ionization chamber in the dedicated holes, the rod created for the ionization chamber was cut in half lengthwise.

Fig. 3 Pictures of the printed bellies over the ATOM phantom and the printed plates (white color) to simulate the 2nd (on the left) and 3rd (on the right) trimesters.

3.2 Chemical composition and physical density

The chemical composition of the resin, determined by the CHONS analyzer is indicated in Table 1.

The remaining proportion unidentified by the CHONS method (0.4%) is attributed to oxygen for the calculated linear attenuation coefficients because the proportion of oxygen was determined subsequently to the other atoms and thus after the calculation of the attenuation coefficients. Thus, the oxygen proportion used was 23,65% which stay inside the associated uncertainties. Sulfur is below the quantification limit (0.91 μg of sulfur for the mass weighed in the analyses).

The density of the resin obtained by the helium pycnometry method was 1.190 ± 0.001 g/cm³.

3.3 Calculated linear attenuation coefficients

Figure 4a presents the calculated total attenuation coefficients (derived from mass attenuation coefficient using the obtained resin density ρ=1.19 g/cm³) and relative differences for the resin compared to water (H₂O), as a function of photon energy. The chemical composition of the resin was taken into consideration, as described previously with 23,65% of oxygen. The relative difference between water and VeroWhite resin attenuation coefficients is within 15% for photon beams delivered by radiotherapy equipment’s (between 0.1 and 10 MeV).

Fig. 4 (a) Calculated total attenuation coefficients of the VeroWhite resin (2.89% N, 65,51% C, 7,95% H and 23,65% O) compared to water (H2O), as a function of photon energy. (b) measured linear attenuation coefficients as a function of RW3 and VeroWhite resin thicknesses for a 6 MV beam.

3.4 Measured linear and mass attenuation coefficients

The values of measured linear attenuation coefficients as a function of RW3 plates and VeroWhite resin thicknesses are presented in Figure 4b. The average relative difference in attenuation coefficients obtained as a function of RW3 and VeroWhite resin thicknesses is about 12% (12,8% when disregarding the 1 cm thickness point measurement).

The mass attenuation coefficients are calculated by dividing by the density of the two materials: 1.19 g/cm³ and 1.045 g/cm³ for the resin and RW3 respectively. The average relative difference obtained between the two materials for the mass attenuation coefficient is 1% (-0.1% when disregarding the 1 cm thickness point measurement).

4 Discussion

The technique used in this study to design and print the desired phantoms allowed for the printing of 3D models with precise geometrical accuracy and smooth opaque surfaces. As a result, the 3D printed structures were placed in contact with the surface of the ATOM® phantom, and the complementary volumes fit together easily. An irregular surface of a printed belly can lead to an air gap with the surface of the phantom affecting the dose measurements. The CT scanner acquisition parameters play a significant role in achieving a good spatial resolution for the 3D printed model to reduce these amounts of air gap and ensure smooth parts that fit well together. The use of a CT scan with a spatial resolution of 0.625 mm was adapted. Moreover, a high-quality 3D printer capable of precise geometric printing, like the one used in this study, is necessary.

The 3D-printed bellies allow for the simulation of two different trimesters of pregnancy in a woman whose morphology is equivalent to that of the ATOM phantom. The definition of these pregnancy bellies was based on gynecological data (fundal height). The fetal dose can thus be easily measured by inserting a FARMER-type ionization chamber into the designated holes according to the stage of pregnancy. These phantoms were developed to assess the influence of certain beam parameters on the fetal dose and are suitable for parametric studies. Using this phantom is well suited for this purpose but not necessarily for managing a real pregnant woman. Indeed, an important factor influencing fetal dose is the distance between the treated volume and the fetus. Assessing the impact of this distance on fetal dose assessment is therefore essential to validate the use of a phantom of fixed size as developed, based on the size of the CIRS female phantom, for a real patient.

The chemical and physical characteristics measured in this study enable a better evaluation of dosimetric equivalences for its use in creating phantoms in the medical field, whether in radiotherapy or radiology. The skeletal measured density of the VeroWhite resin was 1.190 +/−0.001 g/cm³ whereas the manufacturer indicates a density of (1.17-1.18 g/cm³)⁶. An explanation could be that the density provided by the manufacturer considers the air in the volume attributed to the mass, unlike our measurement. Nevertheless, the difference remains small (less than 1%) and confirms that the use of resin-based printing helps to minimize air gaps between layers unlike filament-based methods.

In this study, the dosimetric equivalence of the resin used for printing the bellies was measured and assessed by evaluating linear attenuation. However, these results should be used with a caution: i) the undefined proportion of resin chemical composition, using CHONS analyzer, was attributed to oxygen (0,37%); ii) the experimental conditions for evaluating the attenuation coefficient have some limitations, as described after.

The beam used for the attenuation coefficient evaluation thought measurement is polyenergetic, and its initial spectrum at the accelerator output is not considered. The average energy of a spectrum changes with the traversed thickness due to a variable probability of interaction of photons depending on their energy. Moreover, the use of a beam (3 × 3 cm²) for the measurements produces scattered radiation contributing to the evolution of the spectrum.

The experimental setup assumes that the beam energy is constant regardless of the traversed thickness of RW3 or resin plates. This assumption explains why the evaluated attenuation coefficients are not constant with the traversed thickness. Nevertheless, it should be noted that from approximately 6 cm thickness, the coefficient tends to vary very little with the traversed thickness.

To account for the above limitations in the experimental method, the measured linear attenuation coefficient was compared, under the same conditions, to RW3 taken as reference water equivalent material.

The calculated and measured linear attenuation coefficient for the Verowhite resin agree within 12-15% with water or RW3 plates depending on the approach, with the VeroWhite resin attenuating more than water or equivalent water for radiotherapy energies.

Given the dimensions of the printed bellies, the maximum thickness of resin traversed by the radiation is 10 cm. If the 15% obtained deviation on the attenuation coefficient is applied to the traversed distance, it is like the radiation were passing through 11.5 cm instead of 10 cm. This deviation is deemed acceptable within the scope of our study. Indeed, the bellies were designed to represent the different trimesters, but only 3 specific months of pregnancy were modeled, leading to approximate dimensions for these trimesters. Furthermore, the phantom will be used in the context of a comparative parametric study to evaluate the effects of irradiation parameters on fetal dose, with the goal of providing practical recommendations for managing pregnant women in radiotherapy.

5 Conclusion

In this study, physically realistic bellies representing different trimesters of pregnancy were modeled and then printed using a 3D Printer (Stratasys) and Verowhite resin. To date, this is the first study to characterize the VeroWhite resin for clinical use in radiotherapy. The measured density agrees with the manufacturer’s specifications within 1%. The calculated and measured linear attenuation coefficients for VeroWhite agree within 12-15% with those of water or RW3 plates, within the range of radiotherapy beams. The printed models will be used to assess the impact of irradiation parameters on fetal dose in radiotherapy.

Acknowledgments

The authors have nothing to report. The authors would like to thank the gynecologist for the contributions to this manuscript.

Funding

The authors received no financial support for the research.

Conflicts of interest

The authors declare that they have no conflict of interest.

Data availability statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Author contribution statement

A. Chaikh: methodology, measurements and writing original draft. A.Chaabane: methodology, 3D modeling and printing and measurements. C. Jardin and S. Bassot: assessment of chemical composition and the physical density. T. Beaumont: methodology and workflow. M. Chea, M. Khalal, C. Jenny: methodology and CT acquisition. S. Thomas, C. Huet: methodology, measurements and analysis. A. Isambert, M. Edouard: supervision and writing original draft. All authors contributed to review and editing of the manuscript.

Informed consent

This article does not contain any studies involving human subjects.

References

All Tables

All Figures

	Fig. 1 Representation of the 3D modeling bellies placed over the ATOM® phantom. The fundal heights value and its representation are shown in red, and the green points present the holes used to place the ionization chamber and measured dose.
In the text

	Fig. 2 Experimental setup to measure and compare linear attenuation coefficient using RW3 plates (left) and VeroWhite resin (right) with a 6 MV Elekta beam.
In the text

	Fig. 3 Pictures of the printed bellies over the ATOM phantom and the printed plates (white color) to simulate the 2nd (on the left) and 3rd (on the right) trimesters.
In the text

	Fig. 4 (a) Calculated total attenuation coefficients of the VeroWhite resin (2.89% N, 65,51% C, 7,95% H and 23,65% O) compared to water (H2O), as a function of photon energy. (b) measured linear attenuation coefficients as a function of RW3 and VeroWhite resin thicknesses for a 6 MV beam.
In the text