Evaluation of physical properties and image of polyvinyl chloride as breast tissue equivalence for dual-modality (mammography and ultrasound)

TMP is gradually becoming a fundamental element for quality assurance and control in ionizing and non-ionizing radiation imaging modalities as well as in the development of different techniques. This study aimed to evaluate and obtain polyvinyl chloride tissue mimicking material for dual-modality breast phantoms in mammography and ultrasound. Breast tissue equivalence was evaluated based on X-ray attenuation properties, speed of sound, attenuation, and acoustic impedance. There are six samples of PVC-plasticizer material with variations of PVC concentration and additives. The evaluation of X-ray attenuation was carried out using mammography from 23 to 35 kV, while the acoustic properties were assessed with mode A ultrasound and a transducer frequency of 5 MHz. A breast phantom was created from TMP material with tissue equivalence and was then evaluated using mammography as well as ultrasound to analyze its image quality. The results showed that samples A (PVC 5%, DOP 95%), B (PVC 7%, DOP 93%), C (PVC 10%, DOP 90%), E (PVC 7%, DOP 90%, graphite 3%), and F (PVC 7%, DOP 90%, silicone oil 3%) have the closest equivalent to the ACR breast phantom material with a different range of 0.01–1.39 in the 23–35 kV range. Based on the evaluation of the acoustic properties of ultrasound, A had high similarity to fat tissue with a difference of 0.03 (dB cm− 1 MHz− 1) and 0.07 (106 kg m− 2 s− 1), while B was close to the glandular tissue with a difference of 9.2 m s− 1. Multilayer breast phantom images’ results showed gray levels in mammography and ultrasound modalities. Therefore, this study succeeded in establishing TMP material for mammography and ultrasound. It can also be used for simple quality assurance and control programs.


Introduction
Mammography is the gold method for breast cancer screening, but it has limited sensitivity for detecting lesions in women with dense breast tissue [1]. Meanwhile, ultrasound imaging can distinguish between cysts and solid masses, thereby reducing unnecessary biopsy procedures [2,3]. The combination of these two methods can produce imaging modalities with excellent properties. It is also believed to improve the screening evaluation as well as diagnosis of breast cancer. Several studies have also been carried out on the combination of two or more imaging modalities. Moro et al., 2020 [4] combined Magnetic Resonance Imaging (MRI) and ultrasonography (US) images to improve cervical cancer assessment. Emons et al. 2018 [5] also developed a prototype of a combination between mammography and US images to improve breast cancer detection. Berg et al. 2004 [6] compared the evaluation of breast images from clinical examination using US, mammography, MRI, as well as multimodal and single modality. Furthermore, the result showed that more accurate detection results were obtained using multimodal diagnosis.
The development of a tissue-mimicking phantom (TMP) study is important and challenging due to the importance of multimodality imaging. TMP can be used to assess image quality qualitatively and quantitatively. In mammography and US imaging, it must have an X-ray attenuation coefficient that is equivalent to that of the tissue being imitated [7]. Acoustic properties, such as speed of sound, impedance, and attenuation coefficient must be equivalent to that of the imitated water or tissue [8]. The speed of sound (SoS) and attenuation values recommended by the American Institute of Ultrasound in Medicine (AIUM) are 1540 m s − 1 and 0.3-0.7 (dB cm − 1 MHz − 1 ) [9]. TMP was also recommended to have elasticity, hardness, and Poisson's ratio properties, which are equivalent to the physical parameters of human tissue [10].
TMP materials generally consist of biopolymers and polymers obtained from chemical synthesis. Commercial breast phantoms are available in the market, but they are still limited to a single modality. Several studies have been carried out on the development of TMP materials for multimodalities, such as Gelatin + agar for phantom Positron Emission Tomography (PET), B-Mode US, and elastography [11]. Gelatin + psyllium powder has also been used for US and MRI applications [12], while Polyvinyl Alcohol (PVA) was utilized for US and microwave imaging [13]. However, TMP PVA, agar, and Gelatin could not be used repeatedly due to water evaporation and bacterial growth activity. Gelatin and agar materials had poor elastic properties, which makes them unsuitable for mammography imaging.
An alternative TMP material to overcome these problems is Polyvinyl Chloride (PVC). It is a general synthetic polymer that is easy to synthesize, resistant to bacteria, stable over time, and can be used repeatedly [10]. PVC also has excellent acoustic properties to silicon and polydimethylsiloxane (PDMS) for US imaging. A previous study revealed that PVC with the addition of dioctyl terephthalate (DOP) can have a speed of sound around 1400 ms − 1 [14]. This value is closer to human soft tissue in the range of 1478-1600 ms − 1 [15,16], than silicon which is only about 1000 ms − 1 [17]. Several studies have been carried out on TMP with PVC materials for multimodality applications. He et al. [18] have also used this synthetic polymer for computed tomography (CT) and magnetic resonance imaging (MRI) applications. Based on the elasticity and image quality, the results obtained in CT and MRI modalities require improvement by varying the plasticizer composition and additives. A study by He et al., 2019 [19] used PVC material for multimodality breast phantoms. Furthermore, the results showed that the phantom gave a higher lesion detection rate in mammographic modalities compared to US and MRI. TMP studies for multimodality phantoms only focused on image results and their mechanical parameter. There was also no comprehensive evaluation of tissue equivalence based on X-ray attenuation properties, especially mammography, as well as acoustic properties in the US.
This study focuses on PVC material as a TMP for dual modalities, namely mammography and US, which were evaluated using density parameter, X-ray attenuation, and US acoustic properties, including SoS, attenuation, and impedance. The quality of simple breast phantom images, such as adipose tissue, glandular tissue, and lesions was also investigated. As new imaging techniques and educational tools are developed, this study becomes essential for cost and time-saving quality assurance, as well as control programs in mammography and the US.

Design and fabrication of polyvinyl chloride as TMP
The sample case was a box with an area of 10 × 10 cm² and a thickness of 1 cm. It was produced from Polylactic Acid (PLA) filament using Creality Ender 5 Plus 3D printing (Shenzhen Creality 3D Technology Co., Ltd). Furthermore, TMP was made of polyvinyl chloride (PVC) material and DOP plasticizer with the addition of graphite additives and silicon oil. PVC has an ultrasound velocity of approximately 2330 ms − 1 [20]. Determination of the composition of PVC and DOP was carried out based on the ratio of the mass percentage to the total mass of 200 g, as shown in Table 1. Compositions A -F were prepared by mixing the raw materials in a beaker, followed by constant stirring at 350 rpm and 180 °C. At this temperature, the DOP plasticizer diffused into the PVC particles, followed by stirring to produce a homogeneous matrix characterized by a transparent appearance [21]. After the mixture was cooled to 50 °C, it was poured into the PLA case and cooled to room temperature. The sample fabrication results are presented in Fig. 1(a).

Design and fabrication of a breast phantom for dual modality
The breast phantom was box-shaped with a cross-sectional area of 10 × 10 cm 2 and a thickness of 6 cm. It also consists of two layers, namely the bottom glandular mimicking layer and the top adipose layer. There were characteristic lesions (anomalies) in the form of cylinders between the two layers. Fat and glandular tissue mimicking were made from the composition of TMP A and B samples. Meanwhile, a mixture of gypsum and water in a ratio of 2:1 was used to produce a mimic lesion material. The fabricated breast phantom was made in different layers. The lesion was then inserted into the phantom after the underlayer/glandular creation. It was then coated again with a sample of fat-mimicking material. The results of the fabrication are presented in Fig. 1b.

Measurement of density
The mass density was determined by calculating the massto-volume ratio in units (g cm − 3 ). Furthermore, the sample mass was measured using an electronic digital vernier balance, and the volume was verified with the fluid transfer method.

Measurement of the linear attenuation coefficient
The TMP sample's linear attenuation coefficient was calculated using Lambert Beer's Law by referring to image data with and without objects. Acquisition of image data was carried out using a mammography X-ray machine from 23 to 35 kV. The parameters used include a current of 40 mAs per unit time; Mo/Mo filters, gantry angle of 0°; no compression; and a distance from the source to the object of 65 cm. The image digitization process uses a gray level value with a window level (WL) of 20 and width (WW) of 7. The results were then saved in the DICOM format (.dcm).
The calculation of the linear attenuation coefficient ( µ ) was carried out using Eq. (1).
I 0 is the pixel value obtained from the image without the presence of phantoms or objects, and x is the sample thickness (cm). I is the pixel value obtained from the phantom region image [22,23].

Evaluation of contrast noise to ratio of breast phantoms for dual modality
Contrast noise to ratio (CNR) was calculated with images from inhomogeneous breast phantoms using (4) [25].
where − c in and − c out are the average pixel values of the lesion and breast tissue images. σ 1 and σ 2 are the standard deviations for the same region.

Mass density of TMP
Density values can be used as an inspiration or initial base for assessing the sample, which has the same attenuation properties as the interaction of ionizing radiation in human tissue. The density was used to determine the equivalence of the tissue mimicking the interaction of non-ionizing radiation through the parameter of the acoustic impedance property (on ultrasound). The results of calculating the density of TMP samples made from PVC with a DOP plasticizer are presented in Table 2. In this study, the values of the samples ranged from 0.961 ± 0.012 (sample A) to 1.187 ± 0.035 (g cm − 3 ). The results showed that the concentration of PVC increased along with the density of the tissue-mimicking phantom (TMP).
Ultrasonic Transducer, Sonatest, Cogburn San Antonio, Texas, USA). This system used the pyUn0.py programming, and the raw data was the change in signal amplitude with time. The settings measurement is presented in Fig. 2.
The speed of sound propagation was tested with the pulse-echo method, where the transducer was used as a receiver and transmitter. There was an ultrasonic gel for sample impedance matching between the transducer and sample, hence, signals are sent and a match is received from the pulser/receiver device. The time between peaks was used to determine the speed of sound and was calculated using Eq. (2).
Where d is the distance traveled by the sound wave (2d ) (m), v is the speed of sound m s − 1 , and ∆t is the calculated time difference between two consecutive echo peaks. Furthermore, the determination of sample thickness must be considered properly [24]. The result of multiplying the speed of sound by the density is the acoustic impedance (Z) with a unit (10 6 kg m − 2 s − 1 ) [23].
The acoustic attenuation coefficient was calculated based on the decrease in signal amplitude. The results also showed that the amplitude of the signal reaching the receiving transducer decreases exponentially, and Eq. (3) was used to calculate the attenuation coefficient [24].
where A is the amplitude of the echo signal (mV), A 0 is the amplitude of the pulsed signal (mV), σ is the attenuation coefficient (dB cm − 1 MHz − 1 ), and x is the sample thickness (cm).  Table 3. From Table 3, the attenuation values of the TMP samples that were close to the ACR breast phantom were F, A, B, E, C, and D. Furthermore, they were evaluated for their acoustic properties at ultrasound working frequency for an assessment study of tissue-mimicking breast phantom dual-modality.  [27], who also obtained similar results. The SoS value increased proportionally with the concentration of the material at the same frequency. The addition of a silicone oil dissolving agent (sample F) did not affect the obtained, but the graphite scattering agent had a major influence. Figure 4(b) shows the acoustic impedance (Z) values for each TMP sample with a frequency function. The results show that Z also has a dependence that tends to be linear and constant. Furthermore, its value increased along with the concentration of PVC. The PVC with a 5-20% concentration ranged from 1.47 ± 0.003 to 2.39 ± 0.021 (10 6 kg m − 2 s − 1 ). The results of this Z measurement have a value close to that of Spirou 2005 et al. [28], namely 1.40 ± 0.06 (10 6 kg m − 2 s − 1 ). Figure 5 shows the acoustic attenuation of a TMP sample made of PVC with a variation of DOP plasticizer and additives with a frequency function of 1-5 MHz. The Figure 3 shows the value of the X-ray linear attenuation coefficient for TMP samples made from PVC with a DOP plasticizer. Measurements were carried out by referring to image data in the range of 23-35 kV using a mammography tube. The results showed that the X-ray attenuation level decreased with increasing X-ray energy. This finding indicates that more X-rays were transmitted than absorbed. The attenuation measurements of these TMP samples were compared with those of a commercial breast ACR phantom (www.phantoms.artinis.com/pasmam-constancy). At a  . This shows that the attenuation was related linearly and squarely to the frequency. A 2nd order polynomial fitting function was applied to the acoustic attenuation curve in each TMP sample. This choice was based on the attenuation of water that was suitable using a 2nd order polynomial fitting. In some sample materials, the TMP showed that the attenuation no longer changes linearly, but to a quadratic function for higher frequencies.

Linear attenuation coefficient of TMP
The measurement data were then compared with the actual tissue properties. Sample A here mimics fat tissue as it has a 1.158% difference in mass density, a 2% different X-ray attenuation coefficient compared to the ACR phantom with a 50:50% fat and glandular composition, a 3.96% different speed of sound, a 6.25% different acoustic attenuation coefficient, and a 5% different acoustic impedance. The value of sample A is closer to fat tissue than the other samples. Furthermore, Sample B here mimics glandular tissue because it has differences in mass density, X-ray attenuation coefficient, speed of sound, acoustic attenuation coefficient, and acoustic impedance, respectively 0.002%, 9%, 27.35% (at a frequency of 4.5 MHz), 0.65%. The value of sample B is closer to glandular tissue than the other samples. Based on the evaluation of mass density properties, X-ray attenuation coefficient, as well as acoustics, samples A and B were used as fat and glandular tissue, which mimicked dual-modality breast phantoms (mammography and ultrasound). From the acoustic attenuation coefficients for all PVC samples ranged from 0.51 ± 0.02 to 2.07 ± 0.03 (dB cm − 1 ), at a 1-5 MHz frequency. The values also increased along with the concentration of polyvinyl chloride, and similar results were obtained by Chen et al. 2022 [27]. In this study, the attenuation coefficient of PVC with various concentrations was 1.25 ± 0.162 to 2.07 ± 0.03 dB cm − 1 at a frequency of 5 MHz. These results are consistent with De Carvalho et al. 2016 [29] that the values obtained with an increase in graphite concentration ranged from 0.37 ± 0.01 to 0.61 ± 0.02 (dB cm − 1 MHz − N ). The acoustic attenuation is concentrationdependent. Furthermore, it refers to the loss of mechanical  Furthermore, CNR evaluation of gypsum cylinder lesions and PMMA in mammographic modalities gave values of 7.27 and 3.90, respectively. In the ultrasound modality, the CNR for gypsum lesions near the surface and at the bottom were 9.37 and 21.19, respectively. It was significant because the mass density properties, X-ray attenuation, acoustic ultrasound, and image phantom can provide promising potential for development as a dual phantom -modality in mammography and ultrasound.

Discussion
PVC-Plasticizer DOP material as TMP for mammography and ultrasound dual-modality breast phantom has been evaluated using parameters of density, x-ray attenuation, SoS, acoustic attenuation, impedance, and image. From Table 1, the density of the sample had close tissue equivalent to that of humans. For example, samples A and F were close to fat tissue with a difference of 1.16%. Meanwhile, B and F were close to the breast with a difference of 1.9 and 2.2%, respectively. Samples C, D, and E have good equivalence with the muscle tissues [26]. Based on a previous study, the density obtained was also similar to that of Jeong et al. 2017 [39], where a value of 1.02 (g cm − 3 ) was obtained for PVC-Plasticizer in phantom ultrasound applications. The difference in these values was due to the various concentration levels of PVC. A sample density equivalent to human tissue is the base element for estimating X-ray attenuation properties (mammography), SoS, and acoustic impedance. TMP density close to tissue can have close tissue equivalent to X-ray and ultrasound properties. study of X-rays and sound waves, both of them were close to adipose and glandular tissue. Construction of a box-shaped breast phantom was carried out with a three-layer arrangement and the lesion was then inserted. The bottom to top layer consisted of samples B, A, and B with the insertion of interlayer lesions. The results of dual-modality breast phantom imaging (mammography and ultrasound) are presented in Fig. 6. Figure 6(a) shows the results of breast phantom images that can be formed in mammography modality with an average gray level of 603.40 ± 62.51 at 35 kV. The cylindrical shape of the lesion can be distinguished clearly in the gypsum material, which provided higher contrast than PMMA. The gray-level glandular and fat tissue gave good results even though air bubbles were still formed. Meanwhile, Fig. 6(b) shows the image of the breast phantom in the ultrasound. The cylindrical shape of the lesion was only clearly visible on the gypsum material, and PMMA obtained a dark color. The red ellipse one in Fig. 6a and b shows the same gypsum image (as an anomaly) that can be distinguished in both mammography and ultrasound modalities. The ellipse two in Fig. 6b is images of gypsum at other depths. The second layer obtained a dark color because sample A did not scatter sound waves, hence, a scattering agent was needed. The average gray level and standard deviation obtained from the ultrasound images of the top, middle, and bottom layers were 79.48 ± 20.7, 6.72 ± 1.1, and 26.55 ± 2.16, respectively. Matheo et al. 2018 [38] revealed that the gray level of lactiferous duct, fat, and glandular tissues were 10.18 ± 3.31, 48.41 ± 6.75, and 102.92 ± 11.90, respectively [38]. This study has successfully fabricated a tissue-mimicking of the lactiferous duct and fatty tissue based on the literature data, the glandular part still needs improvement.  [43]. When referring to the parameters of attenuation and acoustic impedance, sample A had an excellent equivalence to fat tissue with a difference of 0.03 (dB cm − 1 MHz − 1 ) and 0.07 (10 6 kg m − 2 s − 1 ), respectively. Sample B also was close to glandular tissue by referring to SoS values and acoustic attenuation. However, the acoustic attenuation parameter still requires further study. Sample A had an excellent close equivalence with the commercial phantom material Zerdine with a difference of 6.7 m s − 1 , -0.03 (dB cm − 1 MHz − 1 ), and − 0.03 (10 6 kg m − 2 s − 1 ), respectively. The composition of the PVC-plasticizer material with agar and PVA materials competed with several studies' results, which are summarized in Table 4.
The difference between TMP samples and human tissue can be caused by several factors, such as the molecular weight of the PVC-Plasticizer, which was low or exceeded during fabrication, thereby affecting X-ray attenuation and SoS, or requiring a scattering agent to increase the acoustic attenuation. Achieving tissue equivalence that satisfies both X-ray and ultrasonic waves is relatively challenging. However, the proposed TMP results were close to breast tissue, including A and B. This indicates that they satisfied the requirements as breast phantom material for dual-modality (mammography and ultrasound). Figure 6 shows breast phantom images from samples A and B with lesions inserted for quality assurance and control. Cylindrical insertion lesions were both seen on mammography and ultrasound modalities. The gray level of breast tissue in the mammography modality obtained results that followed the breast ACR phantom. However, the ultrasound modality on layer A (PVC5%) obtained a gray level close to black. This was due to the lack of a scattering agent to increase the speckle of the ultrasound image. The adipose tissue pulse SoS factor was 1.27% lower than that of glandular tissue. This caused the contribution of the refractive effect to be small for the fat-to-glandular interface, hence, the pulse group's speed difference did not produce artifacts in the image. These findings are consistent with Carvalho et al. [29], where similar results were recorded. The results obtained have provided an important aspect in developing multimodal phantoms. This material has several other advantages, such as low-cost and reusability [38]. Improvements that can be carried out include adding scatter agents and completeness features to perform quality assurance as well as control of mammography and ultrasound modalities.

Conclusion
This study revealed the potential of polyvinyl chloride (PVC) material with the ratio of plasticizer addition for dual-modality breast phantom tissue in mammography and Meanwhile, Fig. 3 shows the linear attenuation coefficients of the X-ray samples that were compared with commercial breast ACR phantoms. The results between samples A, B, C, E, F, and the ACR phantoms gave promising results at 28-35 kV. However, for low voltages, the results were relatively different. From the literature data obtained by Heine et al. 2006 [40], the average X-ray linear attenuation coefficient of all samples has a difference of 1.29 ± 0.25 from that of the 50%-50% glandular-fat tissue [40]. This difference was due to variations in the quality of the X-ray tube from each manufacturer of mammography modalities. Samples A, B, C, E, and F have a close tissue equivalent with the X-ray attenuation properties of the ACR phantom. The value of the linear attenuation coefficient with a voltage obtained a quadratic relationship with a regression value of 0.957 as well as a t-test with a p-value of 0.431, which was greater compared to the set α-value of 0.05. This indicates that the fitting data does not show a significant difference with a regression value of 0.95. These results are consistent with Sato et al.2021 that the linear attenuation coefficient provides a non-linear relationship with X-ray energy [41].  [42] who stated that SoS has a very linear and constant dependence. The σ data showed a quadratic response to frequency variations with an average regression value of 0.978 and obtained a p-value equal to the set α-value of 0.05. The regression results were closer to the value of 1, indicating that the fitting were more accurate. The acoustic attenuation had an increasing trend as the frequency increased. These findings are similar to that of Villa et al. 2020 andZell et al. 2007 [32,42] that the attenuation coefficient increased proportionally with the frequency following the Power Law function.

Conflict of interest
The authors declare that there are no conflicts of interest to disclose.
Ethical approval This study does not involve human participants or animals.

Informed consent
This experiment was carried out on a phantom, and no participation consent was obtained.
ultrasound. Promising results were obtained that PVC-Plasticizer material, including samples A, B, and F had properties close to adipose, glandular, and lactiferous duct tissues based on density parameters, X-ray linear attenuation coefficient with a difference of 0.26-0.55 from the ACR phantom, SoS below 1600 m s − 1 , attenuation at 0.51-0.64 dB cm − 1 MHz − 1 , and acoustic impedance of 1.47-1.64 10 6 kg m − 2 s − 1 . This study succeeded in establishing a generalized breast phantom for quality assurance in mammography and ultrasound modalities. It is also helpful in training, educating, running controlled quality assurance programs, and developing improved dual-modality imaging techniques.
Acknowledgements The authors are grateful to the management of the Gambiran General Hospital, Kediri, Indonesia, and the Medical Instrumentation Laboratory, Engineering Physics Department, Institut Teknologi Bandung, for the permission to conduct this study, as well as to collect and analyze data.