Bioimpedance spectroscopy method for investigating changes to
intracranial dose during transcranial direct current stimulation
Herschel Caytak1 , Isar Nejadgholi2 , Izmail Batkin3 , and Miodrag Bolic
Abstract— Tissue resistance changes upon application of DC
current. We posit that in a similar fashion, that scalp and
skull resistances during trancranial direct current stimulation
(tDCS) are variable, resulting in changes to intracranial dose.
Transcranial magnetic stimulation (TMS), electoencephelogram
(EEG), functional magnetic resonance imaging (fMRI), proton
magnetic resonance spectroscopy (1 H MRS) and functional near
infrared spectroscopy (fNIRS) are technologies used to measure
individual neural reponse to tDCS. These technologies are
complex and may not be directly correlated to intracranial dose.
We therefore present a bioimpedance spectroscopy method
of measuring changes to the intracranial dose in vivo. Scalp
resistance changes are measured during tDCS. Current flow
through the scalp is calculated as the ratio of voltage measured
on the scalp and scalp resistance. Variation of intracranial
current is indirectly calculated from changes in the current
shunted through the scalp. We thus demonstrate a novel
methodology of on-line monitoring of scalp resistance and
current as an objective feedback of estimated individual tDCS
dose.
I. INTRODUCTION
Transcranial direct current stimulation (tDCS), is a non-
invasive technique whereby a low amplitude DC current is
transmitted to the brain via external scalp electrodes [1].
tDCS has shown potential therapeutic benefits for a wide
range of pathologies and social disorders including reduction
of chronic pain, depression, and nicotine craving [1].
One of the main shortcomings of this emerging therapy
is the high variability of tDCS when comparing the effect
of the stimulation on different groups and individuals [2].
Treatment effect has been correlated to intra-cranial dose [3]
which is related to the ratio of scalp/skull impedance [4].
We hypothesize that the intracranial current dose changes
during stimulation due to time-varying resistance changes to
the scalp and skull tissues [5]. Variation of the intracranial
current during stimulation affects the electric field in the
brain and may significantly affect individual response to
tDCS.
*This work was funded in part by Mitacs, NSERC and Nuraleve Inc. We
would like to thank Dr. Adler, Carleton University, for providing us with
the use of the Solartron as well as lab space.
1 Herschel Caytak is with the School of Electrical Engineering and
Computer Science, University of Ottawa, 800 King Edward Ave, Ottawa,
ON, K1N 6N5, hcayt024@uottawa.ca
2 Isar Nejadgholi is with the School of Electrical Engineering and
Computer Science, University of Ottawa, 800 King Edward Ave, Ottawa,
ON, K1N 6N5, i.nejadgholi@gmail.com
3 Izmail Batkin is with the School of Electrical Engineering and Computer
Science, University of Ottawa, 800 King Edward Ave, Ottawa, ON, K1N
6N5, vbatkin@rogers.com
4 Miodrag Bolic is with the School of Electrical Engineering and Com-
puter Science, University of Ottawa, 800 King Edward Ave, Ottawa, ON,
K1N 6N5, mbolic@site.uottawa.ca
978-1-4244-9270-1/15/$31.00 ©2015 IEEE
4
State of the art methodologies used to monitor tDCS re-
sponse include measuring motor evoked potential (MEP) am-
plitude induced by transcranial magnetic stimulation (TMS)
[6], as well as recording changes to electroencephalogram
(EEG) signals [7], functional magnetic resonance imaging
(fMRI) [8], proton magnetic resonance spectroscopy (1 H
MRS) [9] and functional near infrared spectroscopy (fNIRS)
[10]. The majority of these technologies are complex, expen-
sive and involve significant resources to employ. In addition
these methods may not be directly correlated with parameters
associated with intracranial current dose.
We therefore introduce a method of estimating changes to
intracranial dose. A circuit model is developed as a hypo-
thetical description of the time-varying electrical properties
of head tissues in the path of the stimulation current. An
experimental method is then demonstrated to measure in-
vivo change in scalp resistance during tDCS. Concurrent with
tDCS applied using the anodal F3 - cathodal F4 montage
[11], we monitor the scalp impedance using a tetrapolar
arrangement of small electrodes placed between the anode
and cathode. The electrodes are closely spaced to ensure the
measurements are primarily sensitive to the scalp tissue layer.
We use bioimpedance spectroscopy (BIS) measurements to
obtain impedance-frequency spectra of the scalp over the
duration of the stimulation. We fit each impedance spectrum
to the classical Cole model [12] to obtain the DC resistance
of the scalp. The DC voltage drop across the scalp is
monitored during the stimulation. The change in ratio of
voltage to resistance is used to calculate the variation in
the current flow through the scalp and to indirectly estimate
changes to the intracranial dose.
II. TDCS C IRCUIT M ODEL
A. Variable Resistor Model
The head tissues affecting the current path during tDCS
have been modelled as a set of resistors in series and in
parallel [4]. In a similar fashion we model the scalp and the
skull as a network of resistors. A current source provides a
constant current to the network. The stratum corneum (SC
- the outer layer of scalp tissue) resistor is in series with a
parallel combination of scalp (inner scalp tissue below the
SC layer) and skull resistors (Fig. 1). The SC, scalp and skull
are modeled as time varying resistors that describe the rate
of resistance change of each tissue. Following the method of
[4] the brain resistor has been neglected since it is assumed
that the total current passing the skull boundary will enter
the brain.
3448
 by rearranging the terms in eq. 6 we get a decrease of
intracranial dose if
d(Rscalpt )
dt Rscalpt
d(Rskullt )
< (7)
Rskullt
dt
Thus changes to intracranial dose are dependent on the
ratio of the scalp and skull resistances as well as the ratio
of their derivatives.
III. C OLE MODEL
When tissue reactance is plotted in the complex plane
against real resistance, an approximate arc of a circle results,
with intercepts on the real resistance axis at R0 and R∞ . The
Cole Model is an empirical equation that approximates this
characteristic form of impedance spectra plots [12] and is
Fig. 1. tDCS circuit model, the stratum corneum (SC), scalp and skull written as
tissues are modelled as variable resistors.
R0 − R∞
Z(f ) = R∞ + (8)
1 + (jωτ )α
The total time-dependent resistance of the tissues can then
be described as where, Z(f ) is the frequency dependent impedance, ω is
2π ∗ f where f is the measurement frequency in Hz and
Rt = Rsct + Rscalpt ||Rskullt (1) τ is the characteristic time constant of the tissue. α is a
scalar between 0 and 1 that characterizes the phase angle of
or the measurement. Tissue impedance spectra may be fitted to
Rscalpt ∗ Rskullt Eq. 8 using a nonlinear regression algorithm [13] where four
Rt = Rsct + (2) Cole parameters [α, 1/τ, R0 , R∞ ] are calculated to provide
Rscalpt + Rskullt
the closest possible fit to the measured impedance spectra.
where Rsct is the equivalent resistance of the parallel We extract the R0 parameter to calculate the DC resistance
stratum corneum resistors, Rscalpt is the resistance of the of the scalp.
scalp, and Rskullt is the equivalent resistance of the skull
resistors in series. IV. E XPERIMENT
The total current can be described as the sum of the current A. On-line scalp impedance monitoring during tDCS
in the parallel pathways of the scalp and skull or
Our experimental montage is shown in Fig. 2. The study
procedures were approved by the Research Ethics Boards at
Itotal = Iscalp + Iskull (3)
the University of Ottawa and Carleton University, Ottawa,
The proportion of the total current that reaches the brain Ontario, Canada. Written informed consent was obtained
is from the participant in accordance with the Declaration
of Helsinki prior to participation. All assessments were
Ibrain Iskull performed in a controlled laboratory environment. Anodal
= (4)
Itotal Iscalp + Iskull and cathodal 5 ∗ 7cm2 sponge electrodes were secured to
The magnitude of the proportion of the total current that the F3 and F4 locations [11] (defined by the 10 - 20 EEG
reaches the brain is inversely related to the proportional system) on the scalp of a single healthy male volunteer
resistance of the skull. When the resistance changes over age 33. Four small 9 mm diameter electrodes (E21-9S
time this can be written as Disc Electrode) were then applied to the scalp using EEG
paste. The electrodes were spaced at even 1 cm intervals
Ibraint 1 centered on a line between the anode and cathode. The outer
∝ Rskullt
(5)
Itotalt 1 + Rscalp electrodes served as current carrying (CC) electrodes and the
t
inner electrodes were designated as voltage pick up (PU)
I
If Ibraint
totalt
changes over time, the total intracranial dose can electrodes.
be said to be variable. As an example, we set a condition We used the Solatron (1255 Frequency Response Analyzer
Ibrain
d( Itotal
t ) and the 1294 Impedance Interface) to measure impedance
of decreasing intracranial dose where dt < 0. By
t
spectra of the scalp from 1 until 100 kHz logarithmically
deriving the right side of eq. 5 this can be written as spaced with 10 points per decade. The measurements were
obtained using the tetrapolar configuration of EEG elec-
1 d(Rscalpt ) 1 d(Rskullt ) trodes shown in Fig. 2. AC amplitude was fixed at 100
∗ − ∗ <0 (6)
Rscalpt dt Rskullt dt microamps. Three measurements were taken before tDCS

3449
 TABLE I
AVERAGE R0 , PRE , AND DURING T DCS

R0 [Ω] ± SD Number of measurements

Pre tDCS 113 ± .8 3
During tDCS 103 ± 1.1 7

V. R ESULTS
A. Scalp R0 from fit to Cole model
Fig. 3 shows the fit [13] of a baseline impedance mea-
surement to a semi-circle predicted by the Cole model. R0
is the value of the right intercept of the semi-circle and the
real resistance axis.

Fig. 2. Experimental montage; anode F3, cathode F4 used to apply tDCS,

4 small electrodes used for BIS spaced at 1 cm intervals around the midway 10
line, DC voltage is measured between PU1 and PU2. Measurement
Cole Plot
8

−Reactance(ohm)
to ensure the stability and reproducibility of the baseline
scalp impedance. We selected the third measurement as
6
a baseline for comparison with BIS measurements of the
scalp impedance during tDCS.The sponges were hydrated
with 12 ml of 0.9 NaCl solution [14]. TDCS of 1 mA 4
intensity was then applied using the M1000 DC stimulator
(Nuraleve, Ottawa) for 20 minutes. The duration time of
2
each BIS measurement was approximately 2 minutes. Seven
measurements were obtained during the tDCS session with
approximately 1 measurement every 3 minutes. A multimeter 0
(Fluke 289) was used to record the DC voltage between the 90 95 100 105 110 115
Resistance(ohm)
inner PU electrodes.
B. Scalp R0 Fig. 3. Baseline scalp R0 extracted from fitting a plot of the resistance vs
reactance of the measured impedance spectrum to the Cole model
Each impedance spectra measured pre and during tDCS,
was fitted [13] to the Cole model to extract R0t . R0t was
Table I shows the average R0 pre, and during tDCS.
normalized to the baseline measurement obtained pre tDCS
and then plotted vs time where t = 0 is defined as the time of B. Scalp R0 , voltage, current and intracranial current
the first measurement obtained during tDCS (approximately
1 minute after initiation of the stimulation). All on-line
measurements were obtained within 15 minutes from t = 0.
160
C. Calculation of normalized scalp and intracranial current
Normalized Parameters [%]

150
We calculate scalp current over time according to
140
Iscalpt = Vt /Rt (9) Iscalp
130 Vscalp
where Vt is the DC voltage measured between the PU
electrodes and Rt is R0 of the scalp extracted from the Cole Iskull
120
model measured over the experiment time duration. R0scalp
We assume that the small closely spaced electrodes used 110
to obtain BIS and voltage data are sensitive to all the current
that passes through the scalp. We can then calculate Iskullt 100
or the estimated intracranial current according to Eq. 3 as
90
Itotal − Iscalpt = Iskullt (10) 0 5 10 15
tDCS Session Time [Min]
Iscalpt and Iskullt are normalized to Iscalpt=0 and Iskullt=0
respectively so current variation is expressed as proportional Fig. 4. Scalp R0 , voltage, current and skull current change measured
change normalized to the measurement at t = 0 . during pre and during tDCS

3450
 Fig. 4 shows the normalized scalp R0 , scalp DC voltage,
scalp current and estimated intracranial current. R0 is nor-
malized to the baseline scalp resistance before application
of tDCS. Other parameters are normalized to a baseline
measurement taken at the onset of tDCS. R0 is shown to
approach saturation at the beginning of tDCS whereas scalp
voltage and current increase throughout the stimulation.
VI. D ISCUSSION
We modeled the SC, scalp and skull as varying resis-
tors. We experimentally measured the changing scalp DC
resistance using bioimpedance spectroscopy with a tetrapolar
configuration. The advantages of this method is that it
reduces the effect of polarization on the measurement and is
insensitive to electrode contact impedance. Since R0 is fitted
to a theoretical semi-circle, low frequency measurements can
be avoided, allowing for a higher range of probe current
amplitude with potential for an increased measurement SNR.
Table I and Fig. 4 show that R0 quickly saturates at the
onset of stimulation whereas that scalp voltage and current
rises during stimulation but appear to approach a steady state
towards the end of the stimulation. Our measurements show
that at the onset of tDCS, 15% of the total current is shunted
through the scalp, whereas by the end of the stimulation this
proportion rises to 25%.
It is interesting to note that a tDCS modeling study [15]
showed, that for electrodes of the same size and with a
comparable inter-electrode distance, the shunted proportion
of current is around 35% of the total injected current,
more than twice as large as our baseline scalp current
measurement. The discrepancy may be possibly attributed
to a number of limitations of our method. We assumed that
closely spaced electrodes would be mainly sensitive to scalp
impedance, however the impedance of the other tissue layers
such as the skull and brain may significantly affect the global
impedance measurement. In addition, tissue sensitivity is
frequency dependent. The relative contribution of various
tissue layers to the measured impedance may change as a
function of frequency resulting in an inaccurate calculation
of R0 [16]. The estimated proportion of intracranial current
may also be inaccurate since current that penetrates skull may
be shunted through the conductive CSF without reaching
the brain. In addition, we note that the low amplitude AC
current passing through a small cross section of tissue may
limit the sensitivity of the measurement to a local area of
the scalp. A more accurate measurement might be attained
by measuring scalp impedance through an array of closely
spaced electrodes. Since these results are from single subject,
more testing is required to validate the technique.
Despite the sources of error and limitations we mentioned,
we note several important results. Scalp resistance is shown
to be variable at the onset of and during tDCS, resulting
in significant changes to current flowing through the scalp
and into the brain. The variation of scalp and intracranial
current shows that the change in scalp impedance may be a
important factor in determining tDCS intracranial dose.
3451
VII. F UTURE WORK
We present a bioimpedance spectroscopy method for mea-
suring variation of intracranial dose during tDCS. In order to
improve the accuracy of the results, future work will require
modelling the frequency dependence of the various head
tissue impedances as well as new methods for measuring
global scalp properties rather than limited local regions of
the tissue.
R EFERENCES
[1] M. A. Nitsche and W. Paulus, “Transcranial direct current stimulation–
update 2011,” Restorative Neurology and Neuroscience, vol. 29, no. 6,
pp. 463–492, 2011.
[2] J. C. Horvath, O. Carter, and J. D. Forte, “Transcranial direct current
stimulation: five important issues we aren’t discussing (but probably
should be),” Frontiers in Systems Neuroscience, vol. 8, pp. 1–8.
[3] M. Iyer, U. Mattu, J. Grafman, M. Lomarev, S. Sato, and E. Wasser-
mann, “Safety and cognitive effect of frontal dc brain polarization in
healthy individuals,” Neurology, vol. 64, no. 5, pp. 872–875, 2005.
[4] S. Gholami-Boroujeny, A. Mekonnen, I. Batkin, and M. Bolic, “The-
oretical analysis of the effect of temperature on current delivery to
the brain during tdcs,” Brain Stimulation, vol. 8, no. 3, pp. 509–514,
2014.
[5] S. Oh, L. Leung, D. Bommannan, R. Guy, and R. Potts, “Effect of
current, ionic strength and temperature on the electrical properties of
skin,” Journal of Controlled Release, vol. 27, no. 2, pp. 115–125,
1993.
[6] M. Nitsche and W. Paulus, “Excitability changes induced in the human
motor cortex by weak transcranial direct current stimulation,” Journal
of Physiology, vol. 527, no. 3, pp. 633–639, 2000.
[7] P. Schestatsky, L. Morales-Quezada, and F. Fregni, “Simultaneous eeg
monitoring during transcranial direct current stimulation,” Journal of
Visualized Experiments: JoVE, no. 76, pp. 1–11, 2013.
[8] A. Antal, R. Polania, C. Schmidt-Samoa, P. Dechent, and W. Paulus,
“Transcranial direct current stimulation over the primary motor cortex
during fmri,” Neuroimage, vol. 55, no. 2, pp. 590–596, 2011.
[9] M. Rango, F. Cogiamanian, S. Marceglia, B. Barberis, A. Arighi,
P. Biondetti, and A. Priori, “Myoinositol content in the human brain
is modified by transcranial direct current stimulation in a matter of
minutes: A 1h-mrs study,” Magnetic Resonance in Medicine, vol. 60,
no. 4, pp. 782–789, 2008.
[10] A. C. Merzagora, G. Foffani, I. Panyavin, L. Mordillo-Mateos,
J. Aguilar, B. Onaral, and A. Oliviero, “Prefrontal hemodynamic
changes produced by anodal direct current stimulation,” Neuroimage,
vol. 49, no. 3, pp. 2304–2310, 2010.
[11] F. Fregni, P. Liguori, S. Fecteau, M. A. Nitsche, A. Pascual-Leone,
and P. S. Boggio, “Cortical stimulation of the prefrontal cortex with
transcranial direct current stimulation reduces cue-provoked smoking
craving: a randomized, sham-controlled study,” Journal of Clinical
Psychiatry, vol. 69, no. 1, pp. 32–40, 2008.
[12] K. S. Cole, “Permeability and impermeability of cell membranes for
ions,” in Cold Spring Harbor Symposia on Quantitative Biology. Cold
Spring Harbor Laboratory Press, 1940, vol. 8, pp. 110–122.
[13] Y. Yang, W. Ni, Q. Sun, H. Wen, and Z. Teng, “Improved cole
parameter extraction based on the least absolute deviation method,”
Physiological Measurement, vol. 34, no. 10, pp. 1239–1252, 2013.
[14] A. F. DaSilva, M. S. Volz, M. Bikson, and F. Fregni, “Electrode
positioning and montage in transcranial direct current stimulation,”
Journal of Visualized Experiments: JOVE, no. 51, pp. 1–11, 2011.
[15] P. Faria, M. Hallett, and P. C. Miranda, “A finite element analysis
of the effect of electrode area and inter-electrode distance on the
spatial distribution of the current density in tdcs,” Journal of Neural
Engineering, vol. 8, no. 6, pp. 1–11, 2011.
[16] Ø. G. Martinsen, S. Grimnes, and E. Haug, “Measuring depth de-
pends on frequency in electrical skin impedance measurements,” Skin
Research and Technology, vol. 5, no. 3, pp. 179–181, 1999.