IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 35, NO.

4, APRIL 1988 257

Dielectric Properties of Breast Carcinoma and the

Surrounding Tissues
ANDRZEJ J. SUROWIEC, STANISLAW S. STUCHLY, SENIOR MEMBER, IEEE, J. ROBIN BARR,
AND ARVIND SWARUP

Abstract-Relative permittivity of infiltrating breast carcinoma and kHz as compared to the normal or nonmalignant tissues.
the surrounding tissue was measured. The experiments were per- England and Sharples [4] and England [5] examined sev-
formed at frequencies from 20 kHz to 100 MHz at 37°C using an au-
tomatic network analyzer and an end-of-the-line capacitive sensor.
eral samples of breast carcinoma at 10 GHz. Based on
Cole-Cole dielectric parameters were calculated by curve Btting using England et al. [4] and England [5] results, Malard et al.
a computer program. Three main categories of tissues were consid- [6] postulated a possibility of using microwave transmis-
ered: the central part of the tumor, the tumor surrounding tissue, and sion measurements for tumor detection.
the peripheral tissue. Within each category, large spread of the dielec- Roberts et al. [7] observed a fourfold increase of the
tric data for different specimens suggest structural and cellular inho-
mogeneitiesof the tumor tissue. However, certain consistency has been dielectric constant and the loss factor of normal breast tis-
found in the dielectric relaxation time and the coefIieient of the distri- sue and fatty mammary tissues fibrosed by X-rays. Simi-
bution of the relaxation time within each category. The results seem to lar results have been reported by Singh et al. [8] for nor-
indicate that the radio frequency impedance imaging can potentially mal breast tissue and breast tissue containing malignant
be used as a diagnostic modality for the detection of human breast tumors obtained in vivo at low radio frequencies (below
carcinoma.
10 kHz). Recently, Chaudhary et al. [9] studied dielectric
properties of breast carcinoma and normal tissues at fre-
I. INTRODUCTION quencies from 3 MHz to 3 GHz using frequency and time
domain dielectric spectroscopy.
T HE dielectric properties of tumor tissues have been of
interest for several reasons. The dielectric data are
needed in the radio frequency induced hyperthermia as a
We are not aware of any other studies of the dielectric
properties of breast carcinoma at radio frequencies. Also,
modality in cancer treatment as well as in developing the no dielectric measurements have been reported of breast
inhomogenous models simulating the electrical properties carcinoma and the adjacent tissues, i.e., when samples
of the organs with the lesions. Another potential appli- are taken from various parts of the tumor and from the
cation includes radio frequency imaging [11. Besides di- surrounding nonmalignant tissues.
rect utilizations of the dielectric data as indicated above, In this study, we present data on dielectric properties
investigations of the biophysical mechanisms of interac- of infiltrating breast carcinoma and the surrounding tis-
tions of radio frequency fields with tumor tissues may sues in the frequency range from 20 kHz to 100 MHz.
provide some information related to the function-struc-
11. MATERIALMETHODS
ture relationship during the development of the neo-
plasms. A. Tumor Specimens
At radio frequencies, the dielectric properties of bio-
Tumor specimens from seven different patients used in
logical matter are basically determined by the charging of
this study of an exact known history and diagnosis were
the cell membrane through a combined impedance of the
obtained from excisional surgical procedure performed in
intra- and extracellular media. They also depend on the
the Ottawa Civic Hospital. From each specimen, depend-
chemical composition of the tissue [2].
ing of its size, 2-6 samples in form of thin disks (6 mm
Electrical properties of malignant breast tumors have
in diameter and 1 mm thick) were excised. The total num-
been investigated by Fricke and Morse [3]. They found
ber of samples investigated in this study was 28. Sample
significantly higher permittivity of the tumor tissue at 20
preparation procedures have been presented elsewhere,
e.g., Surowiec et al. [lo]. Since the measurements were
Manuscript received October 29, 1986; revised May 22, 1987. This work performed at 37"C, the end-of-the-line sample holder was
was supported by grants from the Ontario Ministry of Labor and the Na-
tural Sciences and Engineering Council of Canada. modified in order to minimize heat losses from the sam-
A. Surowiec is with the Department of Experimental Radiation Oncol- ple, see Fig. 1 . All measurements were completed within
ogy, Bowman Grey School of Medicine, Winston-Salem, NC 27103. 4 h following the surgery. The water contentof the sam-
S. S. Stuchly is with the Department of Electrical Engineering, Univer-
sity of Ottawa, Ottawa, Ont., Canada. ples from different locations in the specimen varied from
J. R. Barr is with the Department of Laboratory Medicine and Cancer 80 to 85 percent. Histological analysis of the samples
Clinic, University of Ottawa, Ottawa Civic Hospital, Ottawa, Ont., Can- showed an infiltrating ductal carcinoma with a focal pap-
ada.
A. Swarup is with Bolriet Technology Inc., Ottawa, Ont., Canada. illary component or a lobular carcinoma. Dash-dot-dash
IEEE Log Number 87 18986. lines in Fig. 2 (A-G) describe the visible boundary of the

0018-9294/88/0400-0257$01.OO 0 1988 IEEE

Authorized licensed use limited to: K J Somaiya College of Engineering - MUMBAI. Downloaded on September 17,2021 at 19:18:43 UTC from IEEE Xplore. Restrictions apply.
 258 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 35. NO. 4, APRIL 1988

Water Bath
A ----1
1T
20

n - , - - N
Pt black electrodes 1TUMOR

Thin mylar tube Glass cyllnder 25 _I

20

Plastlc cylinder
(cortodwith 8 thin maCa Rlm)

Coaxial
Connector
1
Fig. 1. Modified end-of-line coaxial sensor for measurements at temper-
atures different from m m temperature. A metal-coated glass cylinder in
the center conductor and a metal-coated plastic cylinder in the outer con-
ductor form a heat bamer and allow us to maintain samples at temper- C
atures different from the ambient temperature (temperature of the con-
nector). i A

tumor tissue. In some specimens, the tumor formed a hard

core of different density and size.
B. Dielectric Measurements
Dielectric properties of the samples were determined by
measuring the input reflection coefficient of the sample
holder using an automatic network analyzer HP 3577 con-
trolled by an IBM PC computer. For each sample, the 1 25 I
measurements repeated 10-15 times at 101 discrete fre-
quencies evenly distributed in the logarithmic scale in fre-
quency. D
The results at each frequency were averaged. The es-
timated measurement uncertainties were less than 2 per-
cent and 1 percent in the dielectric constant and conduc-
tivity, respectively. In order to minimize the electrode
polarization effect, platinum black electrodes and a sub-
/
stitution technique developed by Schwan [111 were used.
In this technique, the tissue was replaced by a saline so-
lution of the same conductivity and its dielectric proper-
ties were determined at low frequencies. Error correcting
coefficients, obtained by comparing the measured values '10.3-
to the true values of the saline solution, were used to cor-
rect the measured data of the tissues at frequencies below
/
35 2
Fig. 2. The location of the tissue samples excised from the specimens of
100 W Z .
breast carcinoma. Each sample consists of a disk of 6.2 mm in diameter
and 0.8 mm thick, the two numbers indicate: the upper right comer the
111. RESULTSAND DISCUSSION dielectric constants at 100 kHz and 100 MHz, respectively, while the
Fig. 2 shows typical specimens obtained from surgical numbers in the lower left comer indicate conductivities (bold) at the same
frequencies. Specimen's dimensions are in millimeters; thickness 3 and
procedures with cylindrical regions corresponding to the 5 mm. Dash-dot-dash lines describe the visible boundary of the tumor
measured samples. In each region, the numbers indicate: tissue.

Authorized licensed use limited to: K J Somaiya College of Engineering - MUMBAI. Downloaded on September 17,2021 at 19:18:43 UTC from IEEE Xplore. Restrictions apply.
 SUROWIEC et al.: PROPERTIES OF BREAST CARCINOMA AND SURROUNDING TISSUES 259

E
20 >

n
E
0
\
cn
W
E

>-
t
H

>
H

t-
o
3
0
Z

t
0
o

Fig. 3. Dielectric constant of breast carcinoma as a function of frequency.

(*) sample Al, central part of tumor, ( + ) sample A3, tissue surrounding
the tumor ( 0 ) sample C1, mainly fatty tissue containing infiltrating tu-
20
mor cells, ( x ) peripheral sample (G3) located relatively far from central
part of tumor, ( v ) normal (control) breast tissue.

10
'1
10-2 10-1 10 O 10 10
FREQUENCY (MHz)
Fig. 4. Conductivity of breast carcinoma as a function of frequency. ( * )
sample A l , central part of tumor, ( + ) sample A3, tissue surrounding the
tumor, ( 0 ) sample C1, mainly adipose tissue containing infiltrating tu-
L 20 J mor cells, ( x ) peripheral sample ( G 3 ) located approximately 25 mm
Fig. 2. (Continued.) from central part of tumor, ( v ) normal (control) breast tissue.

the upper right comer, the dielectric constants at 100 kHz trating lobular carcinoma, F-infiltrating ductal carci-
and 100 MHz, and the lower left comer corresponding noma, G-infiltrating lobular carcinoma.
values of the conductivities. The dielectric constants and conductivities for a few
Surgical pathology reported the following diagnoses: typical samples are given, as a function of frequency, in
A-infiltrating ductal carcinoma without metastases, B- Figs. 3 and 4, respectively.
infiltrating ductal carcinoma, C-infiltrating ductal car- Our experimental data suggest three types of breast car-
cinoma with focal papillary component, D-infiltrating cinoma. It should be pointed out, however, that within
carcinoma consistent with lobular carcinoma, E-infil- one type one may find samples with widely different prop-

Authorized licensed use limited to: K J Somaiya College of Engineering - MUMBAI. Downloaded on September 17,2021 at 19:18:43 UTC from IEEE Xplore. Restrictions apply.
 260 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 35, NO. 4, APRIL 1988

Fig. 5 . Microscopic view of normal breast tissue and of breast carcinoma:

(a) normal tissue, (b) infiltrating lobular carcinoma, sample E 3 (see lo-
cation, Fig. 2), (c) infiltrating ductal carcinoma, sample F2 (see loca-
tion, Fig. 2), (d) edge of infiltrating ductal carcinoma, sample C1 (see
location, Fig. 2). Tumor cells are seen in the upper part of the specimen.
Also a few tumor cells can be found in the lower part. In the central part,
mostly normal fat cells are observed.

erties probably reflecting various stages of the tumor de- of normally distributed collagen and fat from the sur-
velopment. Thus, the determination of the boundaries be- rounding unaffected breast tissue, (see Fig. 2 , samples A3,
tween different types of times is difficult because of a high A4, B2,03, E 3 , G4,G5,G6).The samples taken from
degree of structural heterogeneity. regions surrounding the center of the tumor are also char-
The first type is representative of the bulk of the tumor acterized by a broad range of ‘dielectric constant ranging
and consists of collagen, elastic fibers, and many tumor from 2.5 X lo3 to 8 X lo3 at 100 kHz. Also the range of
cells. The latter are aggregated into groups, acini and conductivity increments is higher (5-7 ms /cm).
sometimes single cells in a line or column. In these areas, The third type of sample was taken from the periphery,
the malignant cells form the bulk of the specimen, see in other words, from the normal breast tissue distant by
Fig. 2 , (e.g.; samples A l , B1, C3, E2, F1, F 2 ) . These approximately 2 cm from the tumor. This is composed of
samples have low-frequency conductivities between 2 and normal connective tissue and portions of normal glandular
4 mS/cm, with dielectric constants at 100 kHz ranging tissue, see, for example, samples E l , F 3 , G1, G 2 in Fig.
from 2 x lo3 to 6 x lo3. The conductivity increments 2 . The dielectric properties of these samples differ signif-
(the difference between the conductivities at 100 kHz and icantly from the other two categories of tissues. Their di-
100 MHz) are between 4-5 ms/cm for these samples. electric constant is usually less than 500 at 100 kHz, their
The second type of sample has been taken from the in- low-frequency conductivity is below 1 mS /cm and their
filtrating margins of the tumor near the edge of the lesion. conductivity increments are two orders of magnitude
Here, there are fewer tumor cells and a larger proportion smaller than for the two previous tissue categories. The

Authorized licensed use limited to: K J Somaiya College of Engineering - MUMBAI. Downloaded on September 17,2021 at 19:18:43 UTC from IEEE Xplore. Restrictions apply.
 SUROWIEC et al.: PROPERTIES OF BREAST CARCINOMA AND SURROUNDING TISSUES 261

!d)
Fig. 5. (Continued)

dielectric properties of these “normal” tissues are similar should also reflect the structural differences among sam-
to those obtained for adipose tissue by Foster et al. [12]. ples taken from different parts of the tumor. The tissue
In some specimens, see Fig. 2, samples C1,04, El, G3, samples from the tumor core have average dielectric re-
which represent an intermediate stage of tissue, i.e., be- laxation times ( 7 ) between 0.6 and 1.4 ps and static di-
tween normal tissue and the second type of tumor sample, electric constants from 5 x lo3to 9 x io3.~n most cases,
much higher values of the permittivity were observed. the relaxation time of tissues that surround the tumor is
This can be attributed to some tumor cell infiltration pres- considerably shorter (approximately 0.3 ps). The static
ent in these samples, seen for example in Fig. 5(d), that dielectric constant ranged from 3 X lo3 to 5.5 x lo3 for
may contribute significantly to the increase of the dielec- the tissues with shorter relaxation time and from 5 X lo3
tric constant. These differences can potentially be used for to lo3 for the tissue with the longer relaxation time, see
early detection of breast carcinoma. In order to perform Table I. For both categories of tissue samples (central and
further analysis of the experimental data, the dielectric surrounding), a broad distribution of relaxation times
parameters for all samples were calculated using a com- [represented by (a)]was observed, see Table I. Fatty tis-
puter program called STEPIT’ [101, [131, [141 by fitting sues or normal tissues with less tumor involvement (sam-
the data to the Cole-Cole equations [2]. The computed ples F3, G1, and G2) showed narrower distribution of
parameters are shown in Table I. relaxation times, as indicated by lower values of a.
From the previous discussion, one may expect that the Our measurements revealed significant differences in
computed dielectric parameters for the Cole-Cole model dielectric properties between samples taken from different
locations. These differences can be associated with the
‘Indiana University Computer Program Exchange, Department of cellular heterogeneity and structural differences of the
Chemistry, Indiana University, Bloomington IN (D. Chandler, author). tested samples.

Authorized licensed use limited to: K J Somaiya College of Engineering - MUMBAI. Downloaded on September 17,2021 at 19:18:43 UTC from IEEE Xplore. Restrictions apply.
 262 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 35, NO. 4, APRIL 1988

TABLE I
COMPUTER FITTEDDIELECTRIC
PARAMETERS OF BREAST
CARCINOMA AND THE SURROUNDING TISSUES
IN THE FREQUENCY
RANGE
FROM 100 kHz
TO 100 MHz

Dielectric Parameters

Specimen Sample 7 uDC

Location No. 6s cor (P) (Y (ms /cm) Diagnosis

1 6090 50 0.633 0.326 2.20 Infiltrating

2 8260 50 0.595 0.311 3.03 ductal
3 3720 50 0.267 0.283 6.47 carcinoma
4 5170 50 0.288 0.239 6.91 metastases
5 3440 50 0.238 0.238 7.12
6 5710 50 0.263 0.277 6.60
1 3330 50 1.30 0.380 3.3 Infiltrating
2 55 10 30 0.33 0.250 5.70 ductal
carcinoma
1 793 20 0.579 0.232 0.82 Infiltrating
2 482 30 0.603 0.281 0.38 ductal
3 3310 30 0.557 0.310 3.05 carcinoma
1 2740 30 0.269 0.358 3.89 Infiltrating
2 3285 30 0.3202 0.275 5.75 lobular
3 2320 30 0.259 0.278 6.49 carcinoma
4 1253 319 0.670 0.400 3.52
1 3125 167 0.525 0.223 1.02 Infiltrating
2 7060 30 1.605 0.362 4.99 lobular
3 13100 30 1.396 0:349 6.59 carcinoma
4 3080 50 0.45 0.298 3.90
1 8830 30 1.408 0.360 2.78 Infiltrating
2 10170 30 1.210 0.340 2.73 ductal
3 180 10 0.684 0.149 0.120 carcinoma
I 288 10 0.983 0.10 0.161 Infiltrating
2 298 10 1.008 0.075 0.130 lobular
3 675 10 1.391 0.163 0.444 carcinoma
4 7930 50 1.216 0.367 5.33
5 5490 50 0.768 0.379 4.15
6 8660 50 1.415 0.331 5.16

REFERENCES Techniques in Biological Research, W. L. Nastuk Ed. New York:

Academic, 1963, pp. 327-407.
[l] B. D. Sollish, E. H. Frei, E. Hammerman, S. B. Lang, and M. Mo- [12] S. R. Smith and K. R. Foster, “Dielectric properties of low-water
shitzky, “Microprocesser-assisted screening techniques,” Israel J . content tissues,” Phys. Med. Biol., vol. 30, pp. 965-973, 1985.
Med. Sci., vol. 17, pp. 859-864, 1981. [13] R. D. Stoy, K. R. Foster, and H. P. Schwan, “Dielectric properties
[2] K. R. Foster and H. P, Schwan, “Dielectric properties of tissues,” of mammalian tissues from 0.1 to 100 MHz: A summary of recent
in CRC Handbook of Biological Effects of Electromagnetic Fields, data,” Phys. Med. Biol., vol. 27, pp. 501-513, 1982.
Polk and E. Postow Eds. Boca Raton, FL: CRC Press, 1986. [14] A. Surowiec, S. S. Stuchly, and C. Izaguim, “Dielectric properties
[3] H. Fricke and S. Morse, “The electric capacity of tumors of the of human B and T lymphocytes at frequencies from 20 KHz to 100
breast,” J. Cancer Res., vol. 16, pp. 310-376, 1926. MHz,” Phys. Med. Biol., vol. 31, pp. 43-53, 1986.
[4] T. S. England and N. A. Sharples, “Dielectric properties of the hu-
man body in the microwave region of the spectrum,” Nature, vol.
163, pp. 487-488, 1949.
[5] T. S. England, “Dielectric properties of the human body for wave
lengths in the 1-10 cm range,” Nature, vol. 166, pp. 480-481, 1950.
[6] J. R. Mallard and D. G. Lawn, “Dielectric absorption of microwaves Andrzej J. Surowiec was born in Lwow, Poland
in human tissues,” Nature, vol. 213, pp. 28-30, 1967. in 1942. He received the M.Sc. degree in electri-
[7] J. E. Roberts and H. F. Cook, “Microwaves in medical and biolog- cal engineering from the Technical University,
ical research,” Brit. J . Appl. Phys., vol. 3, pp. 33-40, 1952. Gliwice, Poland in 1964 and the Ph.D. degree in
[8] A. Singh, C. W. Smith, and R. Hughes, “In vivo dielectric spec- physics from Silesian University, Katowice, Po-
trometer,” Med. Biol. Eng. Comput., vol. 17, pp. 45-60, 1979. land, in 1972.
[9] S. S. Chaudhary, R. K. Mishra, A. Swarup, and J. M. Thomas, From 1964 to 1973 he was on the staff of the
“Dielectric properties of normal and malignant human breast tissue Silesian Academy of Medicine in the Department
at radiowave and microwaves frequencies,” Indian J . Biochem. Bio- of Medical Physics and Biophysics. In 1972 he
phys., vol. 21, pp. 76-79, 1983. became an Associate Professor in the Academy’s
[lo] A. Surowiec, S. S. Stuchly, and A. Swarup, “Radiofrequency di- Department of Biophysics. He was engaged in
electric properties of animal tissues as a function of time following teaching and researeh in electrical and optical properties of DNA and de-
death,” Phys. Med. Biol., vol. 30, pp. 1131-1141, 1985. sign of electronic systems for electrophysiological studies. From 1973 to
[ 111 H. P. Schwan, Determination of Biological Impedances in Physical 1974 he was a Postdoctoral Fellow in molecular biology at the Centre

Authorized licensed use limited to: K J Somaiya College of Engineering - MUMBAI. Downloaded on September 17,2021 at 19:18:43 UTC from IEEE Xplore. Restrictions apply.
 SUROWIEC et al.: PROPERTIES OF BREAST CARCINOMA AND SURROUNDING TISSUES 263

d’Etude de L’Energie NucMaire in Mol, Belgium. While there, his field tield of radar and microwave technique. From 1959 to 1963 he was with
included research on the interaction of ionizing radiation with DNA and the Warsaw Technical University. In 1963 he joined the UNIPAN, subsid-
genetic manipulation. From 1975 to 1982 he was the Head of the Physics iary of the Polish Academy of Sciences as a manager of the Microwave
Department, Silesian Medical Academy, where he was engaged in studies Instruments Division. In 1970, he immigrated to Canada. From 1970 to
of environmental isotopes in nucleic acids. From 1983 to 1986 he was with 1976 he was with the University of Manitoba, Winnipeg, Canada as an
the University of Ottawa, Canada and carried out research in the areas of Associate Professor in the Department of Agricultural Engineering and
electromagnetic field interaction with animal and human tissues and cells Electrical Engineering. Since 1977 he has been with the Department of
and simulated materials for electromagnetic absorption studies. In 1986 be Electrical Engineering, University of Ottawa, Ottawa, Ont., Canada, where
was with the Department of Physiology and Biophysics, Virginia Com- he is presently a Professor and teaches courses and carries out research in
monwealth University. the field of microwave theory and techniques as well as digital instrumen-
tation and applications of computers. He is the author of over 100 scientific
Damn in the field of microwave theow and techniaues as well as analoa
kldigital instrumentation and holds i8 patents.
-
Stanislaw S. Stuchly (M’70-SM’72) was born
in Lwow, Poland. He received the B.Sc. degree
from the Technical University, Gliwice, Poland
and the M.Sc. degree from the Warsaw Technical
University, both in electrical engineering, in 1953 J. Robin Barr, photograph and biography not available at the time of piub-
and 1958, respectively, and the Ph.D. degree from lication.
the Polish Academy of Sciences, Warsaw, Po-
land, 1968.
From 1953 to 1959 he was a Research Engineer
in the Industrial Institute for Telecommunica- Arvind Swamp, photograph and biography not available at the time of
tions, Warsaw, Poland, engaged in research in the publication.

Authorized licensed use limited to: K J Somaiya College of Engineering - MUMBAI. Downloaded on September 17,2021 at 19:18:43 UTC from IEEE Xplore. Restrictions apply.