Electrical characterization of bolus material as phantom for use in electrical impedance and computed tomography fusion imaging

Electrical impedance tomography (EIT) is a rapidly emerging field in medical imaging. Recently, studies have been conducted on fusing EIT with other imaging modalities like ultrasound [1], computed tomography (CT) [2] etc. The search for a material to be used during testing as phantom is becoming increasingly important, particularly for EIT/CT fusion studies. The required phantom should have electrical conductivity and x-ray absorption coefficient similar to those of in-vivo tissue.

Various studies have been conducted in an attempt to create or improve phantoms to mimic electrical properties of diverse biological tissues at different frequencies [3, 4, 5].

Studies have also been done on evaluating new phantom materials specific to an imaging technique [9] and for different fused imaging techniques [2,10].

A potential phantom material for both CT and EIT is the bolus material, a material widely used for compensating missing tissue as well as modifying radiation doses for skin surface treatment in radiation therapy. Efforts have been made to design bolus materials with mechanical strength and dosimetric properties close to water over diagnostic and therapeutic ranges [6] and to study the function of different materials in dose distribution when used as bolus [7] with recommendation of paraffin for clinical use [8]. The most commonly used standard bolus material is the commercially available bolus with the trade name Superflab [11]. The radiation absorption characteristics of Superflab are equivalent (+/- 2%) to human soft tissue and the density is close to water. However, Gelatin Bolus is also popular as it can be readily prepared at low cost. It has also been tested for density changes overtime and has shown no significant changes, leading to its common use as bolus material. Further, the dielectric properties of gelatin phantoms have been used to simulate biological tissue [12], making it a promising phantom for EIT imaging.

Commonly used phantoms for impedance measurement can be either biological (banana, cucumber) or non-biological (inorganic). As the impedance of biological tissues changes with time, they are not suitable for long-term EIT studies [13]. Further, shape complexity and size variation in biological objects prevents them from being used as suitable phantoms for testing imaging methods. On the other hand inorganic phantoms (Superflab and gelatin) show long-term stability and standardized conductivity [12]. However, both gelatin and Superflab do not show complex impedances as shown by biological tissue. Therefore, these materials can be used as phantoms only in studies based on the resistive component of complex impedance.

Before bolus materials with conductivities similar to biological tissues can be used as phantoms in studies involving fusion of imaging technologies with EIT, the conductive properties of these materials need to be characterized to verify their similarity with in-vivo tissue.

In this paper, we present the conductive characterization of commercially available bolus material Superflab and in-house prepared bolus material, Gelatin Bolus, in frequencies from 10 kHz to 1 MHz. Further, its comparison to in-vivo connective adipose tissue is made and its application in fused EIT/CT imaging is also discussed.

The electrode impedance without the sample, over the frequency range 10 kHz to 1 MHz is shown in Figure 2.

Fig. 2

Figure 3 shows the frequency response of the resistance component of Sample 1: Superflab. The 10 sub-samples were collected from different parts of the same sheet. Frequency response from each sample is represented by an individual trace on the plot.

Fig.3

The mean and standard deviation of the resistance component over the frequency range were calculated, as shown in Figure 4. The x-axis denotes the frequency and the y-axis denotes the mean resistance (bar plot) and standard deviation (line plot) for 10 samples at a 50 different frequencies.

Fig. 4

The frequency response for Sample 2 is shown in Figure 5. Ten samples were also tested to verify phantom homogeneity. Mean and standard deviation of the data are shown in Figure 6.

Fig. 5

Fig. 6

Similarly, the frequency response for in-vivo tissue with mean and standard deviation is shown in Figures 7 and 8.

Fig. 7

Fig. 8

For further comparison, the mean resistivity of Superflab, gelatin, and in-vivo data were plotted simultaneously against frequency sweep, as shown in Figure 9. The x-axis represents the frequency sweep, and the y-axis represents the mean resistivity in ohm cm represented in logarithmic scale.

Fig. 9

As the density of the Gelatin Bolus material does not change over time, though it does experience shrinkage in size due to the evaporation of water molecules, it can be used over long periods of time following proper storage procedures. The procedure consists of storing the material wrapped in saran wrap in a refrigerator. Similarly, the conductive properties of the gelatin bolus have been studied over a time period of 12 months, to verify its usability for EIT based multi-modality imaging over such a long period of time. Figure 9 shows the impedance response of the Gelatin Bolus over time. Three samples of same size have been considered for comparison. All samples were created six months apart. One sample was only one day old and considered freshly prepared.

The effect of temperature on the conductivity of the phantom was also studied from 20ºC to 26ºC temperature. Previously, all the readings were taken at room temperature (20ºC) for all the samples, whereas the tissue from which in-vivo data was collected had a higher temperature close to 37ºC (body temperature).

The Gelatin Bolus temperature was increased above room temperature to 26ºC (beyond this temperature the bolus melts). It was found that no significant change occurred in response to increased temperature.

Electrical impedance measurements are based upon tissue specific electric field distribution and the measurability of currents or potentials on the surface of the region of interest. At low frequencies the current is mostly extracellular, whereas with increasing frequencies intracellular contributions become increasingly significant, and the observations become more cell specific [17]. Thus, the tissue impedance is frequency dependent. Tissue electrical impedance is different at different frequencies depending upon the tissue classification. Therefore, phantoms to be used in EIT should show frequency dependent impedance behavior similar to in-vivo tissue.

Comparative studies, over biological (banana, cucumber) and non-biological phantoms (homogenous saline phantom with acrylic plastic, stainless steel), show that the complex conductivities of biological objects change more with time and therefore cannot be reused as imaging objects for EIT based studies [13]. Further, biological objects are also not good as phantoms for CT imaging due to their irregular shapes and sizes. Therefore, the required EIT phantoms should show long-term stability and standardized conductivity over a frequency range to be useful for comparative experimental studies.

Interestingly, the non biological phantom made with gelatin, simulates biological tissue well in terms of dielectric properties and is also popular for electrical impedance imaging studies [12]. Gelatin consists of mobile ions: Amino, hydroxyl, and thiol groups are present with multitude of hydrogen ions and the small chain fragments can readily migrate through the gel under the influence of a small field [18].

Superflab and Gelatin Bolus are both recommended for use in radiological applications. For CT-EIT fusion imaging, specifically, the frequency dependant impedance of the sample needs to be compared with in-vivo collected data. Few constraints need to be added to overcome the effect of high impedance of skin that is observed below the 1000 Hz range and the effect of electrode polarization at low frequencies [19]. Keeping these constraints in mind the comparative frequency range considered in the present study was from 10 kHz to 1 MHz. Electrode impedance without sample present was also collected over the entire frequency range. Figure 2 depicts a resistance of 260 Ω from 10 kHz to 100 kHz, and thereafter it starts increasing. Therefore, the most suitable frequency range to use for comparative study was from 10 kHz to 100 kHz.

Unfortunately, gelatin based phantoms do not show complex impedance whereas biological tissue show complex impedance. Consequently, the present study is capable of comparing only the real part of the complex impedance of biological tissue to bolus material. In the future, complex impedance behavior materials need to be studied for better representation of in-vivo tissue in terms of complex electrical properties.

Sample 1 (Superflab) showed a lot of inhomogenity and variability in the real impedance value throughout the range of frequencies tested, as shown in Figure 3. Therefore, though Superflab is one of the most popular materials to be used as bolus material in radiation imaging, it is not suitable for phantom development in multi-modality imaging involving EIT. Figure 4 is very useful in further elaborating the variability and inconsistency of the sample, as observed the standard deviation is very high. The difference between in-vivo tissue resistivity and superflab sample is quite high (Figure 9) making the sample unsuitable as a phantom for validation in EIT.

The frequency response obtained using Sample 2 (Gelatin Bolus) was very consistent, as shown in Figure 5 and 6, showing much smaller standard deviation compared to Superflab. As shown in Figure 9, though the frequency response of the Gelatin Bolus is similar to in-vivo collected data, the resistivity of Gelatin Bolus is higher. The conductivity of the Gelatin Bolus can further be increased by addition of conducting ions e.g. NaCl. This may lead to a much closer response to the in-vivo tissue response. Therefore, Gelatin Bolus can be used in multi-modality imaging involving EIT in the frequency range from 10 kHz to 1 MHz. It also has the advantage of being easily prepared in-house.

With passage of time the resistance of Gelatin Bolus decreases due to loss of water molecules, as shown in Figure 10. Freshly prepared gelatin bolus has response closest to in-vivo data.

Fig. 10

The CT images showed a CT number with mean 17.2 and standard deviation of 12 for sample 1 (Superflab) and mean of 116.7 and standard deviation of 5 for sample 2.

Gelatin Bolus has similar frequency response but lower conductivity compared to in-vivo adipose connective tissue at frequencies from 10 kHz to 1 MHz. A freshly prepared Gelatin Bolus has much closer response to in-vivo tissue as compared to samples stored over 6 months. After heating the same sample to 26ºC we observed no significant change in electrical properties compared to sample stored at room temperature. Superflab bolus is not recommended for EIT/CT fusion studies as its electrical properties are quite different from biological tissue. Therefore, freshly prepared gelatin bolus shows the closest similarity to in-vivo tissue in the 10 kHz to 1 MHz frequency range.