INSTITUTE OF PHYSICS PUBLISHING
PHYSICS IN MEDICINE AND BIOLOGY
Phys. Med. Biol. 50 (2005) 4245–4258
Tissue-mimicking phantom materials for narrowband
and ultrawideband microwave applications
Mariya Lazebnik1, Ernest L Madsen2, Gary R Frank2
and Susan C Hagness1
1Department of Electrical and Computer Engineering, University of Wisconsin, Madison,
WI 53705, USA
2Department of Medical Physics, University of Wisconsin, Madison, WI 53706, USA
Received 29 April 2005, in final form 24 June 2005
Published 31 August 2005
Online at stacks.iop.org/PMB/50/4245
We propose and characterize oil-in-gelatin dispersions that approximate the
dispersive dielectric properties of a variety of human soft tissues over the
microwave frequency range from 500 MHz to 20 GHz.
are mimicked by selection of an appropriate concentration of oil.
materials possess long-term stability and can be employed in heterogeneous
configurations without change in geometry or dielectric properties due
to osmotic effects.Thus, these materials can be used to construct
heterogeneous phantoms, including anthropomorphic types, for narrowband
and ultrawideband microwave technologies, such as breast cancer detection
and imaging systems.
Traditionally, phantoms that closely mimic the physical properties of various human tissues
have been invaluable for the development and testing of medical imaging modalities such
as ultrasound, magnetic resonance imaging (MRI), computed tomography (CT) and others
(see, for example, Fong et al (2001), Madsen et al (1982a), Madsen et al (1988) and Surry
et al (2004)). Over the past several decades, with the rapid expansion of cellular phones
and other personal devices that emit electromagnetic energy at microwave frequencies, as
well as the use of electromagnetic energy for medical imaging and therapy, there has been
growing interest and research on the interactions of electromagnetic waves with biological
tissues, particularly at microwave frequencies. Therefore, the need for biological phantoms
that mimic the electromagnetic properties of tissues at these high frequencies has greatly
0031-9155/05/184245+14$30.00© 2005 IOP Publishing LtdPrinted in the UK4245
4246M Lazebnik et al
Many different tissue-mimicking (TM) phantom materials have been developed to
simulate the properties of high-water-content and low-water-content biological tissues at
discrete frequencies. These materials can be roughly divided into several classes, according
to the base ingredient. Although a comprehensive literature review is beyond the scope of this
paper, several such classes will be briefly discussed.
One such material, developed by Guy (1971) for simulating high-water content tissues,
consists of TX-150 (a gelling agent, widely referred to as ‘super stuff’), polyethylene powder,
water and sodium chloride (NaCl). By varying the relative proportions of the ingredients,
high-water-content tissues can be simulated at frequencies between 13.56 and 2450 MHz
(Chou et al 1984). Unfortunately, this material does not yield a useable low-water-content
tissue simulant (Lagendijk and Nilsson 1985). Guy (1971) also proposed a material consisting
of Laminac 4110 (a polyester resin), acetylene black and aluminium powder to simulate bone
and fat around 100–1000 MHz. Cheung and Koopman (1976) adapted both materials to
simulate human tissues at 8.5 and 10 GHz.
Another type of material, described first by Bini et al (1984) and subsequently by
Andreuccetti et al (1988), uses polyacrylamide gel as the chief ingredient. These materials are
attractive due to their optical transparency and gel-like mechanical properties. However, the
resulting phantoms have very limited shelf life—several hours when exposed to air or several
weeks when kept in an air-tight container (Bini et al 1984). More recently, these materials
were adapted by Surowiec et al (1992), McCann et al (2003), Davidson and Sherar (2003),
and others. These materials involve complicated fabrication methods as well as chemicals that
may be difficult to obtain.
Gelatin-based materials are attractive due to their stable mechanical properties and ease
of fabrication. Marchal et al (1989) described a water–gelatin recipe to simulate high-water-
content tissues from 10 to 50 MHz. Materials of different dielectric properties are constructed
by varying the gelatin concentration. However, these materials do not allow for heterogeneous
phantoms that are stable over a long period of time because of solute or solvent diffusion that
occurs when two materials with different gelatin concentrations are placed in direct contact.
to simulate human skin. Again, since the gelatin concentration is varied to achieve different
dielectric properties, this formulation does not allow for stable heterogeneous phantoms.
In addition, several other materials have been described. For example, Lagendijk and
Nilsson (1985) used a ‘dough’ (consisting of flour, oil and saline) to model low-water-content
tissues at 451 MHz. Robinson et al (1991) created materials for simulating muscle and fat
at 1000 MHz. The muscle simulant consisted of ethanediol, water, salt and gelatin, while
the fat simulant was composed of ethanediol, gelatin and polyethylene powder. Nikawa et al
(1996) described a material that utilized silicone rubber with carbon fibre. By varying the
conductivities could be achieved. The use of silicone rubber allowed for a simple fabrication
narrowband spectra for which they were designed, they would be difficult to use in a wideband
application since a different material composition would be required for every frequency of
interest (Chou et al 1984, Andreuccetti et al 1988, Robinson et al 1991). Therefore, additional
work has been done on creating broadband tissue simulants. For example, Surowiec et al
(1992) succeeded in creating a single polyacrylamide gel material covering the frequency
range between 500 MHz and 3 GHz. Youngs et al (2002) described the quantitative design
the use of percolative conductor–insulator composites for this application. Chang et al (2000)
Tissue-mimicking phantom materials 4247
created a solid conductive plastic that simulated the dielectric properties of muscle from
300 to 900 MHz. These materials consisted of polyethyl methacrylate and carbon black and
could be cast into any arbitrary shape.
In this paper, we address the need for TM materials that enable the construction of
stable heterogeneous and anthropomorphic phantoms for ultrawideband (UWB) applications.
One such emerging UWB technology is the detection of breast cancer using low-power UWB
2001, Fear et al 2002, Bond et al 2003, Hernandez-Lopez et al 2003, Nilavalan et al 2003,
El-Shenawee 2004, Huo et al 2004, Li et al 2004, Davis et al 2005). Promising theoretical
results and preliminary laboratory experiments have brought this technology to the point of
initial pre-clinical studies. Anthropomorphic breast phantoms are needed to bridge the gap
between the relatively simple phantoms that have been used in experimental feasibility studies
to date and the much more complex environment of the actual human breast to be encountered
in pre-clinical studies. This UWB application serves as a compelling illustration for the
rationale behind the development of the TM materials presented in this paper. Since the breast
is a heterogeneous tissue volume with constituents possessing dielectric properties that span
the biological spectrum, the specifications for TM materials dictated by this UWB application
are of a general nature. For example, the development of breast tissue phantoms requires
TM materials for low-water-content tissues (breast fat), higher-water-content tissues (breast
glandular tissues and cancerous lesions) and tissues of intermediate water content (skin). In
addition, it is necessary that different TM materials can be juxtaposed without the risk of their
properties changing due to diffusion across the interface.
To satisfy the above requirements, we have created a TM phantom material over the
frequency range of 500 MHz to 20 GHz that approximates the dispersive dielectric properties
of various biological tissues. Similar materials have previously been developed for ultrasound
phantoms (Madsen et al 1982d). These gelatin-based materials contain varying percentages
of a solution of 50% kerosene and 50% safflower oil, which allows materials with a wide
range of dielectric properties to be constructed. A critical property of these materials is the
ability to create heterogeneous and anthropomorphic configurations with long-term stability
of mechanical and electromagnetic properties. In addition, these materials are elastic as well
as inexpensive and easily fabricated and moulded.
The remainder of this paper is organized as follows. Section 2 discusses the method for
producing the TM materials as well as the technique for measuring their dielectric properties.
Section 3 presents the dielectric properties of these materials over a broad frequency range,
as well as a comparison between the dielectric properties of the TM materials and the
experimentally determined dielectric properties of biological tissues.
2. Materials and methods
2.1. Method of production of the TM materials
First, an aqueous gelatin solution and a solution of 50% kerosene and 50% safflower oil
(henceforth referred to as ‘oil’) are produced and brought to a temperature of 50◦C. Then
quantities of gelatin solution and oil are combined and a surfactant added. Vigorous stirring
creates a uniform emulsion. The emulsion is cooled to 40◦C, and an amount of formaldehyde
solution is added. Then the emulsion is cooled to about 34◦C and poured into a container such
as a mould. The gelatin matrix surrounding the oil droplets congeals as the temperature drops
below 26◦C.Within24hchemical cross-linkingof the gelatinmolecules bythe formaldehyde
raises the melting point of the gelatin matrix to over 100◦C.
4248 M Lazebnik et al
The following is the detailed procedure for producing an oil-in-gelatin dispersion having
a volume per cent oil between 0 and 50%.
(i) In a small beaker mix 0.2 g of p-toluic acid (powder) into 10 ml of n-propanol. Heat
with stirring until the p-toluic acid has dissolved.
(ii) Mix the solution of p-toluic acid and n-propanol into 190 ml of 18 M cm deionized
water in a Pyrex beaker, the water being at room temperature.
(iii) Add—with stirring—34 g (dry mass) of 200 bloom gelatin derived from calfskin
(Vyse Gelatin Company, Schiller Park, IL, USA) to the mixture produced in (ii); the
mixture should still be at room temperature and the gelatin granules should be wetted.
(iv) Cover the beaker with a plastic film, such as Saran WrapR ?(polyvinylidene chloride,
The Dow Chemical Company, Midland, MI, USA), held in place with a rubber band
and heat the mixture in a double boiler. (The hot water jacket should surround the
sides and bottom of the beaker to prevent local overheating of the mixture.)
(v) When the mixture has become transparent (at about 90◦C) and no air bubbles are
suspended beneath the surface, remove the beaker from the double boiler.
(vi) Stir the mixture to produce uniformity and then remove any air bubbles from the
(vii) Partially immerse the beaker in a bath of cool water (20–25◦C) and stir the mixture
until it cools to 50◦C.
(viii) Remove the beaker from the water bath and pour 200 ml into another beaker which
already contains a quantity of oil at 50◦C. If a volume per cent oil of 50% is desired,
then the amount of oil should be 200 ml.
(ix) Vigorously stir the mixture with a spoon bent at a right angle at its base until most oil
droplets are less than about 0.2 mm in diameter (see Madsen et al (1982b) for a more
(x) Continuingtostirvigorouslyandusingasyringe, add0.56mlof‘UltraIvory’(Procter
and Gamble, Cincinnati, OH, USA) liquid surfactant or its equivalent3per ml of oil.
The emulsion should become uniform and—for larger percentages oil—nearly white.
stirring. (Vigorous stirring should no longer be necessary to maintain uniformity.)
(xii) Using a needle and syringe add 2.16 g of formaldehyde solution (37% by weight
formaldehyde—e.g., Catalog no. F-79, Fisher Scientific, Pittsburgh, PA, USA). The
amount of formaldehyde solution is 0.0108 g per ml of the gelatin solution.
(xiii) Cool in the water bath to about 34◦C and pour into the container or mould for cooling
and solidifying. Allow at least 5 days for formaldehyde cross-linking of gelatin to be
If a volume per cent oil of greater than 50% (up to 80%) is desired, then (viii) of the
(xiv) Remove the beaker from the water bath and pour 200 ml into another beaker which
already contains 200 ml of oil at 50◦C.
(xv) Continuing to stir vigorously and using a syringe with a needle, add 0.56 ml of ‘Ultra
Ivory’ (Procter and Gamble, Cincinnati, OH, USA) liquid surfactant or its equivalent
per ml of oil in step 8. (For 200 ml oil, add 11.2 ml of surfactant.) The emulsion
should become uniform and—for larger percentages oil—nearly white.
(xvi) Continuing to stir vigorously and without adding any more surfactant, add about 100
ml of 50◦C oil and stir vigorously until the emulsion has again become uniform. (Do
not add any more surfactant.)
3Other dish washing detergents should suffice, though we have not investigated them.
Tissue-mimicking phantom materials 4249
TM material #1
TM material #2
Figure 1. Diagram of the ‘hockey-puck’ sample, used to test the stability of the TM materials’
dielectric properties when two different samples are placed next to each other.
Table 1. Compositions of tissue-mimicking materials. Abbreviations: p-tol. acid: p-toluic acid,
n-prop: n-propanol, formald.: formaldehyde, surf.: surfactant, wt: weight.
(xvii) Add another 100 ml of 50◦C oil with vigorous stirring until the emulsion has again
(xviii) Repeat step (xvii) until all oil needed has been added and the emulsion is uniform.
In table 1 are shown the percentages by weight of all components in four different
materials, corresponding to 10%, 30%, 50% and 80% (volume per cents) oil.
2.2. Production of samples for dielectric properties measurements
Two types of samples were produced for the purpose of dielectric properties characterization.
We generated homogeneous samples of the TM materials by pouring 20 ml of molten material
(see step (xiii) in section 2.1) into small glass jars (diameter = 3.8 cm and height =
4 cm). The jars are covered to prevent desiccation and left undisturbed while congealing
occurs. The flat upper surface that forms is then accessed during the dielectric spectroscopy
materials can be formed in direct contact with no change in geometry or dielectric properties
with time. For these tests, a cylinder (dubbed the ‘hockey-puck’ sample) was produced with
one half occupied by a material containing 50% oil and the other half occupied by a material
containing 0% oil. These materials were in direct contact with each other. See figure 1 for
a schematic diagram of the ‘hockey-puck’ container. First, a cylindrical container was made
from an acrylic ring with diameter of 7.6 cm, length 2.5 cm and wall thickness 6 mm; a 1 cm