Professional Documents
Culture Documents
AbstractTissue electrical impedance is a function of its ar- to lung cancer [1]. The 5-year relative survival rate for all
chitecture and has been used to differentiate normal and cancer cancers found while still confined to the prostate approaches
tissues in a variety of organs including breast, cervix, skin, and 100% due to the effectiveness of treatment procedures such as
bladder. This paper investigates the possibility of differentiating
normal and malignant prostate tissue using bioimpedance spectra. brachytherapy and radical prostatectomy which are typically
A probe was designed to measure impedance spectra over the administered in these situations to prevent metastasis [1].
range of 10 kHz to 1 MHz. The probe was fully characterized using Detecting prostate tumors early is, therefore, beneficial to in-
discrete loads and saline solutions of different concentrations. creasing the total survival rate associated with their occurrence.
Impedance spectra of five ex vivo prostates were measured in the The two most widely accepted screening methods for
operating room immediately following radical prostatectomy.
Wilcoxon signed-rank tests were used to compare the normal and prostate cancer are serologic measurement of prostate-specific
malignant findings. The impedance probe had a signal-to-noise antigen (PSA) concentrations and digital rectal examination.
ratio (SNR) 84 dB across the entire spectrum and measured Although capable of detecting a high percentage of cancers,
a tissue volume of approximately 46 mm3 . At 10 kHz, prostate these screening methods are limited [2]. Cancer screening using
conductivity ( ) ranged from 0.232 S/m to 0.310 S/m for tumor and
measurements of PSA concentrations has led to a dramatic
from 0.238 S/m to 0.901 S/m for normal tissue. At 1 MHz the ranges
were 0.301 S/m to 0.488 S/m for tumor and 0.337 S/m to 1.149 S/m increase in incidence rates and concomitantly radical local
for normal. Prostate permittivity ( r ) ranged from 6.64 104 to therapies [3], however there is no clear evidence screening
1.25 105 for tumor and from 9.08 104 to 4.49 105 for normal increases survival rates [4], [5]. In addition, overdiagnosis as
tissues at 10 kHz. And, at 1 MHz the r ranges were 9.23 102 a result of screening has contributed to an increased number
to 1.88 103 for tumor and 1.16 103 to 2.18 103 for normal
tissue. Both and r of tumor tissue were found to be significantly
of procedures that are perhaps unnecessary since a large pro-
lower than that of normal tissue (P 0 0001). Conductivity and portion of these cancers may never reach a stage of clinical
permittivity are both higher in normal prostate tissues than they significance [6], [7]. Although the clinical value of the method
are in malignant tissue making them suitable parameters for tissue is still being debated, the sensitivity and specificity of PSA
differentiation. This is in agreement with trends observed in other screening have been reported to be in the ranges of 70%80%
tissues reported in much of the literature. Expanded studies are
needed to further validate this finding and to explore the biological and 60%70% , respectively [8]. Digital rectal examinations
mechanism responsible for generating the results. is limited to detecting palpable cancers in the posterior and
lateral aspects of the gland, where only 85% of cancers arise
Index TermsCancer screening, electrical impedance spec-
troscopy, prostate cancer, tissue electrical properties. [9]. Sensitivity and specificity of digital rectal examinations
have been reported to be 59% and 94%, respectively [10].
Bioimpedance is a property related to a tissues resistance to
I. INTRODUCTION electrical current flow and its ability to store electrical charge.
It is predominantly a function of tissue architecture including
P ROSTATE cancer is the most widely diagnosed malig-
nancy in males. In 2005, 232 090 new cases and 30 350
deaths were attributed to the disease, which was second only
cellular size and density, cellular spacing, and the constituents of
the extracellular matrix (ECM). Impedance differences between
tissue types may be markers for pathologic processes such as
cancer. In fact, Skourou et al. [11] have shown that a tissues
Manuscript received April 28, 2006; revised October 21, 2006. This work was
supported in part by the National Institute of Health (NIH) and National Cancer
impedance signature may be more sensitive to the presence of
Institute under Grant NIH-NCI PPG P01CA080139-06A2. Asterisk indicates tumor tissue than conventional imaging techniques of computed
corresponding author. tomography (CT) and ultrasound. Use of electrical impedance
*R. J. Halter is with the is Thayer School of Engineering, Dartmouth College, signatures to differentiate normal and neoplastic states has been
Hanover, NH 03755 USA (e-mail: ryan.halter@dartmouth.edu).
A. Hartov is with the Thayer School of Engineering, Dartmouth College, reported in studies of cervical [12], breast [13], skin [14], and
Hanover, NH 03755 USA. He is also with the Norris Cotton Cancer Center, bladder tissues [15].
Lebanon, NH 03766 USA. Prostate cancers are typically found in the peripheral zone of
J. A. Heaney is with the Department of Urology, Dartmouth-Hitchcock Med-
ical Center, Norris Cotton Cancer Center, Lebanon, NH 03766 USA. the gland and constitute a dense arrangement of cancerous ep-
K. D. Paulsen is with the Thayer School of Engineering, Dartmouth College, ithelial cells often in the form of small, infiltrating glands with
Hanover, NH 03755 USA. He is also with the Dartmouth Hitchcock Medical a proportionate decrease in stromal volume [16]. These mor-
Center, Lebanon, NH 03766 USA, and Norris Cotton Cancer Center, Lebanon,
NH 03766 USA.
phological differences between normal and malignant prostate
A. R. Schned is with the Department of Pathology, Dartmouth-Hitchcock tissue along with the known sensitivity of electrical impedance
Medical Center, Lebanon, NH 03766 USA. spectra to cellular structure prompted Lee et al. [17] to consider
Color versions of one or more of the figures in this paper are available online using tissue bioimpedance as a measure for distinguishing the
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TBME.2007.897331 disease. In a small series of 6 subjects, the Lee study showed that
Fig. 3. Conductivity and permittivity of different concentrations of saline solution. Measurements are in agreement with Gabriels published values [19].
Fig. 4. TS measurements. (a) Measured impedance as a function of probe distance from bottom of saline tank (0.07 M NaCl). Vertical line shows the TS associated
with a change of less than 0.1%. (b) TS of the impedance probe in different concentrations of saline solution.
Probe signal-to-noise ratio (SNR) was determined by ob- M NaCl. In addition, the impedance spectrum of a volunteers
taining a set of 100 measurements with a 1 load resistor tongue is shown. The impedance values are comparable to those
connected between the inner and outer conducting elements reported by Gabriel et al. [19] in which a different type of probe
of the probe. The SNR was found to vary from 89.6 dB at 10 was used.
kHz to 84.9 dB at 1 MHz, which corresponds to a noise level
of 0.0033% and 0.0056% at the two frequencies, respectively. IV. PROBE SAMPLING VOLUME
Further testing with discrete resistors verified that the calibra- We use probe sensitivity to describe the volume of tissue
tion procedure removed the parasitic impedances properly and that influences the impedance measurements. Radial sensitivity
the measured impedance values were accurate. Fig. 2 shows (RS) is defined as the area parallel to the probe surface that in-
the magnitude and phase of impedances measured with discrete fluences the data; it includes the area directly under the probe as
resistors of 740 , 1 , and 2.7 . The phase shift is load well as the region outside the probe edge where fringing effects
dependent and for impedances equivalent to tissue ( ) is may occur. Transverse sensitivity (TS) describes the depth along
less than 1 at 1 MHz. the axial direction of the probe that influences the impedance re-
The probe geometry factor was determined experimentally sult. Final probe sensitivity is defined as the product of the area
by measuring the capacitance, , at 100 kHz while the probe tip of RS and depth of TS.
was submerged in DI water. A central frequency of 100 kHz was RS and TS were both measured experimentally by inserting
chosen to limit the effect of polarization which occurs at lower the probe in a beaker of saline and translating the probe from
frequencies [18] and minimize the small parasitic phase shifts the edge of the beaker towards the center (RS measure) and
occurring at the highest frequencies. Impedance measurements from the bottom of the beaker upwards (TS measure). The edge
were subsequently recorded for saline solutions of varying NaCl and bottom of the beaker represent nonconductive boundaries
concentrations and the geometry factor was used in conjunc- preventing the flow of current. Impedance measurements varied
tion with (1)(4) to calculate the conductivity and permittivity by less than 0.5% as the probe was translated inward from the
spectrum of each solution. The impedance spectra are shown edge of the beaker, demonstrating that the fringing effects are
in Fig. 3 for concentrations ranging from 0.01 M NaCl to 0.09 minimal and that the field lines are contained within the area
1324 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 54, NO. 7, JULY 2007
Fig. 6. Histological slide of prostate tissue from case PR002. Normal tissue is
visible on the left side of the image and prostatic adenocarcinoma dominates the
right side.
TABLE I
PATIENT CASE PATHOLOGY
Fig. 7. RX spectra of both normal and tumor sites for all five cases. Circles represent the minimum jXj value and are the impedances at the notch frequency for
each case.
Fig. 8. Averaged conductivity and relative permittivity spectra for tumor and normal prostate tissues. Dashed lines signify normal tissue and solid lines signify
tumor tissue.
The conductivity difference between tissue types (normal No significant correlations were found between the histolog-
versus tumor) was found to increase with increased signal ical assessments and measured impedance data due to the lim-
frequency from an average of 0.046 S/m at 10 kHz to 0.078 ited number of samples collected and the approximate co-local-
S/m at 1 MHz. In contrast, permittivity differences were found ization between the impedance and histological measurements.
to be much larger at lower frequencies, having an average
difference of 37 387.3 at 10 kHz and 67.8 at 1 MHz. Wilcoxon VII. DISCUSSION
signed-rank tests were used to compare the differences in data The observed increase in conductivity and decrease in
between tumor and contralateral sides for full spectral data sets permittivity with increasing frequency (Fig. 8) is consistent
. When the entire frequency spectrum was analyzed with reports for other tissues [20]. At low signal frequencies,
both and of tumor tissue were found to be significantly polarized cellular membranes act as charge storage devices
lower than that of normal tissue . (high ) and barriers to ionic conduction (low ). As signal
1326 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 54, NO. 7, JULY 2007
TABLE II
MEASURED IMPEDANCE RANGES AT EACH DECADE
TABLE III
COMPARISON OF DATA TO LEE et al. [17]
diagnostic for differentiating tumor from normal tissue within modalities for enhanced tissue differentiation or may be used
an individual patient. The results from this study and Lee et al. in conjunction with a trans-rectal ultrasonic probe to aid in
suggest a possible threshold of 0.8 for the ratio of conductivities. locating areas of suspicion during prostate biopsy procedures.
An expanded study is, however, necessary to determine how sen-
sitive and specific bioimpedance would be using this threshold. REFERENCES
Although this investigation includes only five subjects, the [1] Cancer Facts & Figures Am. Cancer Soc., Atlanta, GA, 2005, pp. 160.
trends are repeatable and suggest that there is a significant dif- [2] H. Carter, Prostate cancers in men with low PSA levelsmust we find
them?, New Engl. J. Med., vol. 350, p. 2292, 2004.
ference in the impedance measured between normal and tumor [3] SEER Cancer Statistics Review, 19731999 Nat. Cancer Inst.,
prostate tissue. It is well established that cancer causes morpho- Bethesda, MD, 2002.
logical changes in tissue, and that different impedance spectra [4] P. Carroll et al., Prostate-specific antigen best practice policyPart I:
Early detection and diagnosis of prostate cancer, Urology, vol. 57, p.
stem from these variations. In the benign state, prostate tissue 217, 2001.
consists of ducts lined by two layers of epithelial cells. Between [5] Polascik, T. J. Oesterling, and A. Partin, Prostate specific antigen: A
decade of discoverywhat we have learned and where we are going,
these ducts is a large volume of fibromuscular stroma making J. Urol., vol. 162, p. 293, 1999.
up the ECM. Embedded in the ECM is extracellular fluid which [6] T. Stamey, The era of serum prostate specific antigen as a marker for
acts as a conduit for current flow. At the onset of tumor growth, biopsy of the prostate and detecting prostate cancer is now over in the
USA, in Br. J. Urol. Int., 2004, vol. 94, p. 963.
the ECM is infiltrated with proliferating neoplastic cells where [7] T. Stamey et al., The prostate specific antigen era in the united states
the tumorous glands become smaller in size and more densely is over for prostate cancer: What happened in the last 20 years?, J.
packed than normal glands, and the stroma is reduced because Urol., vol. 172, p. 1297, 2004.
[8] M. Brawer, Prostate-specific antigen: Current status, Cancer J. for
of the increase in cancerous epithelial cells. The loss of stroma Clinicians, vol. 49, p. 264, 1999.
and increase in epithelial density decreases the volume of ex- [9] J. Epstein, Pathology of prostatic neoplasia, in Campbells Urol., P.
tracellular fluid in which the current flows and may result in Walsh, Ed., 8th ed. Philadelphia: Saunders, 2002.
[10] A. Hoogendam, F. Buntinx, and H. de Vet, The diagnostic value of
the observed increase in tissue impedance. This mechanism of digital rectal examination in primary care screening for prostate cancer:
tumor growth is common in other types of tissue. A meta-analysis, Family Practice, vol. 16, p. 621, 1999.
[11] C. Skourou et al., Feasibility studies of electrical impedance spec-
Further studies incorporating histological measures such as troscopy for early tumor detection in rats, Physiological Meas., vol.
epithelia-to-stroma ratios, tissue water content, and cellular size 25, pp. 335346, 2004.
as a function of the tissue impedance may help to elucidate the [12] B. Brown et al., Relation between tissue structure and imposed elec-
trical current flow in cervical neoplasia, Lancet, vol. 355, pp. 892895,
subtleties of why neoplastic prostate tissue has been found to 2000.
have lower conductivity relative to its normal counterpart which [13] J. Jossinet, Variability of impedivity in normal and pathological breast
may help with further tissue differentiation [24]. Given the lim- tissue, Med. Biol. Eng. Comput., vol. 34, pp. 346350, 1996.
[14] P. Aberg et al., Skin cancer identification using multifrequency elec-
ited number of studies investigating prostate bioimpedance it trical impedance A potential screening tool, IEEE Trans. Biomed.
would be beneficial to measure a larger sample of prostates and Eng., vol. 51, no. 12, pp. 20972102, Dec. 2004.
more importantly make measurements in conjunction with the [15] B. Wilkinson et al., Electrical impedance spectroscopy and the di-
agnosis of bladder pathology: A pilot study, J. Urol., vol. 168, pp.
post-resection procedure where the probe placement and his- 15631567, 2002.
tology can be precisely coregistered. For the normal side of pa- [16] M. Ernstoff, J. Heaney, and R. Pechek, Eds., Prostate Cancer Blackwell
tient PR005, a high impedance was measured. After further ex- Science Oxford, U.K., 1998.
[17] B. Lee et al., Bioimpedance: Novel use of a minimally invasive tech-
amining the associated histological slide, the tissue in that area nique for cancer localization in the intact prostate, Prostate, vol. 39,
was found to have a mixture of cystic atrophy, inflammation, pp. 213218, 1999.
[18] H. Schwan, Determination of biological impedances, in Physical
glandular hyperplasia, and high grade prostatic intraepithelial Techniques in Biological Research, W. Nastuk, Ed. New York:
neoplasia (PIN). The histological analysis, however, considered Academic, 1963.
a tissue area larger than the probe sensitivity and better co-lo- [19] S. Gabriel, R. Lau, and C. Gabriel, The dielectric properties of bio-
logical tissues: II. Measurements in the frequency range 10 Hz to 20
calization of the probe placement with respect to the area of his- GHz, Phys. Med. Biol., vol. 41, pp. 22512269, 1996.
tological analysis may help to isolate the pathological state of [20] C. Gabriel, S. Gabriel, and E. Corthout, The dielectric properties of
the tissue influencing these impedance measurements. biological tissues: I. Literature survey, Phys. Med. Biol., vol. 41, pp.
22312249, 1996.
[21] J. da Silva, J. de Sa, and J. Jossinet, Classification of breast tissue by
VIII. CONCLUSION electrical impedance spectroscopy, Med. Biol. Eng. Comput., vol. 38,
pp. 2630, 2000.
A probe used to measure the electrical bioimpedance of [22] L. Sha, E. Ward, and B. Stroy, A review of dielectric properties of
prostate tissue was constructed, validated, and characterized normal and malignant breast tissue, Proc. IEEE, 2002.
for spectroscopic recordings from 10 kHz to 1 MHz. Each [23] J. Jossinet and M. Schmitt, A Review of Parameters for the Bioelec-
trical Characterization of Breast Tissue, Ann. NY Acad. Sci., pp. 3041,
measured prostate exhibited significantly higher impedance 1999.
values for tumor relative to normal tissue. Our data is consis- [24] W. Wells et al., Analysis of the microvasculature and tissue type ratios
in normal vs. benign and malignant breast tissue, Analytical Quanti-
tent with prior reports in the literature, with the added benefit tative Cytol. Histol., vol. 26, no. 3, pp. 166174, 2004.
that our results are not probe specific. It is envisioned that a
bioimpedance probe may be incorporated into typical screening Authors photographs and biographies not available at the time of publication.