This article has been accepted for publication in IEEE Sensors Journal.

This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/JSEN.2023.3341168

IEEE SENSORS JOURNAL, VOL. XX, NO. XX, XXXX 2017 1

UWB Microwave Functional Brain Activity

Extraction for Parkinson’s Disease Monitoring
Youness Akazzim , César Palacios Arias ,Student Member, IEEE, Marc Jofre , Otman EL Mrabet
,Member, IEEE, Jordi Romeu , Fellow, IEEE, and Luis Jofre-Roca , Life Fellow, IEEE

Abstract— Microwaves have proven their imaging

capabilities to visualize the body composition for
medical applications thanks to their penetration
inside biological structures. In this context, this
paper presents a novel methodology that aims to
extract not just the internal morphology but also
the brain’s functional activity using the UWB Pulse
Amplitude Modulation (PAM) technique to have
simultaneously functional monitoring and imaging
capability and apply it to monitor the Parkinson’s
disease. The radiofrequency system is composed
of two orthogonal sets of double UWB probes op-
erating in the frequency range of 0.5 GHz - 1.5 GHz.
An experimental set-up has been devised that
avoids complex in-vivo testing, albeit allows a sys-
tem proof-of-concept validation. A bio-tag consist-
ing of an optically modulated photodiode is used
to emulate local medium changes associated to
cell activity. The proposed system is used to first
extract the modeled brain Action Potential (AP) to
validate the performance of the bio-tag, and then to
monitor the Parkinson’s disease (PD) based on the
beta frequency band character within basal gan-
glia–thalamocortical (BGTC) which is a key marker
for the PD. The results show a good capability of
locating and differentiating the signals generated
within the phantom by the bio-tag, alternatively
emulating the healthy and PD’s state, based on the
frequency. The obtained results of the functional monitoring technique on distinguishing the healthy from non-healthy
brain model activity, as well as in the phantom mimicking the average proprieties of a human head, will serve as a basis
for detecting functional diseases in the future.
Index Terms— UWB Functional Imaging, Action Potential, UWB Pulse Amplitude Modulation, Functional Diseases,
Parkinson’s Disease.

I. I NTRODUCTION

The use of microwaves in medical applications is a promis-

This work was financially supported in part by CICYT PID2019- ing solution for human body monitoring, thanks to its advan-
107885GB-C31, PID2022-136869NB-C31, 2017 SGR 219, (FI-
SDUR)(2021 FISDU 00195), PDR-2014-2022/56-30157-2021-2A grant
tages in terms of innocuity, portability, and cost. Based on the
of the Generalitat de Catalunya, and Prueba de Concepto PDC2022- difference in electrical properties of the internal composition
133091-I00 grants of the Ministerio de Ciencia, Innovación y Universi- of the human body tissues, numerous studies have been
dades (Spain).
Y. Akazzim, C. Palacios Arias, J. Romeu and L. Jofre-Roca are with
conducted for microwave imaging [1]–[7], sensing [8]–[10],
the Dept. Signal Theory and Communications, Technical University of or functional signal extraction of the brain or the heart [11]–
Catalonia, Barcelona 08034, Spain. [14] using single frequency analysis.
M.Jofre is with the Dept. Network Engineering, Technical University of
Catalonia, Barcelona 08034, Spain. Due to the progressive death of neurons with age, neurode-
Y. Akazzim, O.El Morabet, are with System of Information and generative or functional pathologies such as Parkinson’s (PD),
Telecommunications Laboratory (LaSIT), FS, Abdelmalek Essaadi Uni- or Alzheimer’s (AD) diseases [15], [16] have become one
versity, Tetouan, Morocco.
Corresponding author: Y. Akazzim (email: of the most spread illnesses. The World Health Organization
[email protected] (WHO) reports that PD has doubled in the last 25 years, and

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in IEEE Sensors Journal. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/JSEN.2023.3341168
2 IEEE SENSORS JOURNAL, VOL. XX, NO. XX, XXXX 2017

the global estimates show over 8.5 million individuals with as the circuital and electromagnetic analysis of the bio-tag
PD (with an increase of 81% since 2000) and caused 329000 used to generate the functional signals. In Section III, the
deaths [17]. numerical and experimental validation is conducted to prove
The main cause of Parkinson’s disease (PD) is the loss the capability of the system to detect a functional activity such
of dopamine in the midbrain substantia nigra [18], which as the Action Potential (AP) within the human head-brain
leads to a progressive neurodegenerative disorder. The PD model. In section IV, an experimental analysis of the beta
symptoms start appearing after the loss of more than 60%, of frequency band oscillation changes for differentiating between
the dopaminergic neurons, or 80% of dopamine concentration two signals representing the healthy and PD’s state based
in the putamen [19]–[21]. There are treatments available for on the firing rate frequency is discussed. Finally, preliminary
PD based on medicines such as Levodopa which provides conclusions are presented in Section V.
the brain with dopamine [22], or surgical as Deep Brain
Stimulation (DBS) [23], [24], but just applied in advanced II. F UNCTIONAL M ICROWAVE S YSTEM
stages of the illness.
Diagnosis methods for neurodegenerative diseases, some of
them initially intended for cancer detection, are Functional CPU
Magnetic Resonance Imaging (fMRI), Computed Tomography RF Coaxial cabling
FSG Control cabling
scans (CT scans), or Positron Emission Tomography scans VNA
AC
(PET scans). These techniques have drawbacks such as pa-
tients must undergo numerous tests and scans, additionally, RF-S
DC4
these approaches may be expensive, uncomfortable, or incon- DC3
DC2 Cylindrical volume approaching
clusive. DC1 the human head
For the early-stage diagnosis and monitoring of PD, re- Two set of TX/RX
BT
searchers explore specific markers that help on understanding double probe

the progression of the disease for early treatment. These 400mm

markers can be clinical as tremors [25], biochemical as neural

density [26], prodromal as rapid eye movement (REM) [27], z
450mm 400mm
y
pathophysiological as the loss of the dopaminergic neurons x
which decreases gradually in the substancia nigra [28] and
affects the firing rate of the action potential [29], or the Fig. 1. The functional monitoring system, where the transmitting and
receiving RF links are selected by the Arduino Control (AC) to measure
changes in the beta frequency band oscillatory character that the functional changes of the Bio-Tag (BT) within the tank of liquid
happens in the basal ganglia–thalamocortical (BGTC) [14], mimicking the human head-brain medium.
[30]–[32]. The functional signal extraction within the brain
has been studied for different contact-electrode applications In this section, the setup used to validate the proposed
using a single frequency technique as for the brain rat [11], the method for microwave functional disease’s is presented in
human brain [13], and recently in the basal ganglia to monitor Fig.1, which consisting into two main subsystems:
Parkinson’s disease [14], but clearly different for the extraction • The UWB imaging system: containing an orthogonal set
of the brain activity in terms of both the wireless contact- of two EGRH probes [1], [37], sequentially connected to
less character and the focusing capability of the microwave a TX/RX vector network analyzer collecting the 4 × 4
technique proposed in this work. UWB scattering parameters.
In this paper, we propose a novel method that combines lo- • The functional signal modeling unit (bio-tag): composed
cating the origin of the signal related to the functional disease by a photodiode illuminated through an optical fiber by
using UWB imaging, and extracting these specific electrical a LED [38] emitting a Pinc blue light power.
activities using the UWB pulse amplitude modulation [33],
These two subsystems are connected to a computer controlling
[34] in order to differentiate the healthy from the parkinsonism
the different instruments, and processing the measured S-
brain based on the beta frequency band oscillatory character.
parameters to locate the bio-tag, and to extract the functional
The radiating system is composed of four Extended Gap Ridge
signals.
Horn (EGRH) antennas [1], distributed in two orthogonal sets
of double UWB probes. A low RF disturbing bio-tag element,
based on a GHz operation response photodiode, is first used A. The UWB Imaging Subsystem
to insert a millisecond time-varying signal, to validate the The UWB imaging system in Fig.1 contains the RF Tx/Rx
system capability by extracting the brain spikes up to kHz addressing circuitry and the four radiating elements (probes).
frequencies [35], [36] within a medium mimicking the human The RF subsystem sequentially addresses (through a combina-
head permittivity, this bio-tag is then used as the origin of the tion of two RF-Switches (RF-S) and four directional couplers
frequency band oscillatory within basal ganglia to produce the (DCi, i=1,2,3 and 4) controlled by an Arduino (UNO) unit) the
frequency signal of 20Hz to mimic the healthy brain and 15Hz Tx and Rx ports of the Vector Network Analyzer (VNA) ZNB
for the parkinsonism one [14]. 40 R&S to each one of the 4 UWB EGRH probes emitting a
The remainder of this paper is organized as follows. Section power of PEGRH = 10dBm to the head model. The system
II presents the proposed functional imaging system, as well is located inside a liquid tank with overall dimensions of

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in IEEE Sensors Journal. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/JSEN.2023.3341168
AUTHOR et al.: PREPARATION OF PAPERS FOR IEEE TRANSACTIONS AND JOURNALS (FEBRUARY 2017) 3

450 × 400 × 400 mm3 , filled with liquid material mimicking point-value in the order of (0.1 nA, 0 V ) for the “rest” and
the human head-brain medium of permittivity εr ≃ 57.0 and (1 mA, 100 mV ) for the “active” states respectively [43], [44].
ei
σ ≃ 0.6 S/m [39], forming an optimized imaging geometry In this way, when a cell is illuminated at a frequency, fmw we
of two 90-degrees separation double-probe sets [1] forming a will recover a “structural non-modulated” signal at the same
virtual cylinder in black that has a diameter of dbrn
phn = 200
ei
fmw frequency corresponding to the volumetric cell scattering
mm approaching the average size of a simplified human and a “transmembrane modulated” signal at the frequency
sg ei
head. The RF subsystem is immersed in the liquid and the flf ± fmw corresponding to the scattering produced by the
container to have simultaneously the virtual head model and changing membrane. Our interest will focus on recovering
the matching liquid inside the probe antennas to minimize the this modulated signal because it is the component carrying the
reflections from the contour of the phantom. information on the membrane’s functional state. As elaborated
In order to protect the photodiode, it is covered with a in [44] for a frequency of 1 GHz the ratio between the two
thin plastic layer that also keeps the optical fiber oriented field components corresponding to the “membrane-modulated”
to the active zone of the photodiode, and it is placed in a and to the “structural-non-modulated” will be in the order of
position shifted from the equivalent center of the brain which −90 dB that will require high-sensitivity (high dynamic range)
corresponds to the one pair of the substantia nigra [40]. detection systems. In order to produce the modulating signals
A Control Processing Unit (CPU), is a computer that con- mimicking the brain functional activity as mentioned above
nects all the hardware parts. It contains a dual-core processor (i.e. Parkinson-like signals) the combination of a photodiode
with 64GB of RAM that processes the results in less than FDS015 from Thorlabs [38] with an active area of 10 × 10
6 minutes, and sends the necessary commands to the (VNA) µm2 and overall size of 20.0 × 4.7 × 6.4 mm3 connected
and Arduino Control (AC) units, to collect the set of complex through an optical fiber to a LED modulator M455F3 [45] is
(amplitude and phase) measured scattering signals sending used. The unit is fed by a 2000 series Picoscope generator able
in parallel the necessary commands to the Functional Signal to generate the different signals (the ON PD electric state to
Generator (FSG) to produce the different functional activity simulate the “active biological state” and the OFF PD electrical
signals in the bio-tag. state to simulate the “resting biological state”).
Based on its equivalent circuit [38] (the parameters in Table
I) the ON and OFF values of its corresponding impedance
B. Equivalence of the Photodiode with the Cellular
were obtained. The data has been extracted for the center
Membrane Behavior
frequency fcn = 1 GHz of the operating frequency band of
Cellular, as well as intracellular, membranes exhibit a dis- the EGRH probes 0.5 GHz - 1.5 GHz. In the case of the
tinct non-linear electrical behavior due to the potential barrier photodiode driven with a Ibt = 0.8 mA which corresponds to
resulting from the difference between the inner and outer the maximum incident power Pinc = 5 mW , the simulated
electrolytes and the action of ion-pumps [41]. In the absence of ON
complex impedance (ON state) is ZL1 ≃ (31.7 − j4.2) Ω.
an applied electromagnetic field, the transmembrane potential When the photodiode is OFF (dark current 0.03 nA), the
difference ∆ϕ is equal to the cell “resting” potential V0 OF F
equivalent impedance is ZL2 ≃ (0.6−j244.8) Ω. The values
(≃ −100 mV for a typical cell) [42]. When a cell enters of the current generated at the photodiode for both electrical
into an active membrane potential state, it generates itself ON (bio-active state) of 0.8 mA and electrical OFF (bio-rest
a low frequency (in the order of up to kHz) transmem- state) of 0.03 nA approach quite well the currents passing
brane voltage excess potential ∆ϕ = V0 + δϕ (resulting in through the cellular membrane mentioned above of 1 mA and
∆ϕ = 40 mV ). On top of the self-generated low-frequency 0.1 nA, respectively, that guarantees similar radiated values
voltage, the illuminating external microwave signal like the for the cellular membrane and the photodiode.
one analyzed in this work (the cell is relatively sensitive
to these electromagnetic frequencies) adds a high-frequency TABLE I
small-signal transmembrane voltage excess potential on the T HE PARAMETERS OF THE P HOTODIODE M ODEL
cell, which is proportional to the electric field applied. As a Parameter Equation (resource) Value
result, a transmembrane current density J⃗m (having the two Bio-tag current Ibt = Pinc × Rλ
components corresponding to the self-generated low-frequency Ibt Responsivity of FDS015 Rλ = 0.16 A/W 0.8 mA
sg Incident power Pinc = 5 mW
signal flf and to the externally induced microwave signal ID Dark current from datasheet [38] 0.03 nA
ei
fmw ) is generated. The current-voltage response of the mem- CJ Junction Capacitance from datasheet [38] 0.65 pF
brane is known to be fairly well approximated by a nonlinear Rsh Shunt resistance from datasheet [38] 100 kOhm
diode-like relationship of the form J⃗m = J0 (e(δϕ/VT ) − 1)
with typical values in the order of J0 ≃ 10−5 A/cm2 and
VT ≃ 5 mV [43]. As results of the non-linear behavior of
the cellular membrane, a combined modulated signal will C. Electromagnetic Analysis of the bio-tag
appear carrying the specific information of the membrane state As previously mentioned, the bio-tag when illuminated by
sg ei
flf ± fmw . For the case of the specific cellular functional ac- an incident RF signal (as in Fig.2) will be responsible for
tivities considered here, the active regions may be approached approaching the effect of the cells creating the low-frequency
by a circular surface with diameter in the order of 10 µm functional signals AP, resulting in the modulating effect pro-
to 50 µm result in transmembrane current-voltage (Im , δϕ) duced by the non-linear behavior of the cell membrane, and

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in IEEE Sensors Journal. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/JSEN.2023.3341168
4 IEEE SENSORS JOURNAL, VOL. XX, NO. XX, XXXX 2017

the differential scattering field is calculated analytically using

mod
(3) and results in ∆S21 ≃ −92 dB. To detect the variation
of equivalent impedance produced by the bio-tag, we need a
dynamic range larger than −90 dB (in the order of 120 dB),
compatible with the sensitivity levels of the actual network
analyzer.

D. Functional Signals Generation

In this section, we validate the bio-tag on producing
functional signals within the head-brain medium based on
the equivalent impedance change of the photodiode. The
Fig. 2. The modulated back-scattered signal scenario.
interaction process between the brain nerves can be seen as
chemical as for the neurotransmitters such as the dopamine
finally reradiating back the RF modulated signal. This bio- in the Synaptic Cleft, or equivalently electrical as for the
tag may be modeled as the combination of a modulating (AP) based on different membrane potential in the axons [49].
ON OF F
photodiode with impedances ZL1 and ZL2 and a radiating
elements formed by its actual leads. In this way, when the
bio-tag is illuminated by the RF incident field radiated by
the corresponding Tx probe the induced RF currents into the
photodiode produces a modulated reradiated (scattered) signal
due to the non-linear effect of the diode, that is significantly
below the non-modulated component due to structural radia-
tion of the photodiode, similarly to what it will happen into the
real biological cell where the membrane modulated scattered
signal is significantly below the scattering produced by the
whole cell.
In the following, the basic electromagnetic expressions of
the scattering effect are presented. The photodiode with its
leads is first modeled as a dipole of length lbt (20 mm,
in the order of a half-wave dipole in the brain medium at Fig. 3. The experimentally generated AP using the bio-tag.
the operating band) loaded alternatively with a modulating
ON OF F The experimental operation of the AP has been conducted
impedance (ZL1 , ZL2 ).
hw by illuminating the bio-tag with a blue light LED oriented
The equivalent impedance of the photodiode’s leads Zbt is
calculated using (1) [46] as: to the active zone of the photodiode through an optical fiber
controlled by the signal generator Picoscope 2000 series,
lbt /λbrn
phn − 0.45
 
hw approaching the shape of the AP. The voltage values for
Zbt = 73 + j43 Ω (1)
0.05 the two states of the photodiode measured with the high
impedance oscilloscope are VbtON ≃ 140 mV when it is ON,
where λbrn
phn is the wavelength within the head-brain medium. and VbtON ≃ 0.2 mV when it is OFF, with ∆Vbt ≃ 140 mV ,
The differential scattered field (difference between the mod-
close to the values of the cell membrane. The real-time signal
ulated and the non-modulated scattered fields) produced by the
generated by the bio-tag to model the AP is presented in
bio-tag can be obtained by (2) as in [47], [48]:
Fig.3, which has a frequency in the order of fbt ≃ 1 kHz,
(λbrn hw 2
phn Gbt ) which mimics the firing rate of a realistic brain AP spike -
∆σbt = |ρL1 − ρL1 |2 (2) 1ms duration- [50].
4π
Z ON −(Z hw )∗ Z OF F −(Z hw )∗
where ρL1 = Z L1
ON
bt
hw and ρL2 = Z L2
OF F +Z hw , and
bt
III. THE NUMERICAL AND EXPERIMENTAL
L1 +Zbt L2 bt
Ghw
bt is the antenna gain of the photodiode leads modeled as VALIDATION
a half-wave dipole. This section presents the validation of the proposed func-
The modulated part of the back-scattered signal from the tional microwave system, which is the combination of two
bio-tag at the receiving probe can be written as: operations: locating the origin of the targeted functional
(λbrn 2 2 activity by using the UWB RF signals and extracting the
mod phn ) Gp ∆σbt brn rx
∆S21 = rx )4 exp(−2αphn rbt ) (3) functional activity parameters (firing rate and shape) from the
(4π)3 (rbt
back-scattered modulated signals using the pulse amplitude
brn
where Gp is the gain of the probe antenna, and αphn is the demodulation.
attenuation in the human head-brain medium. An initial analysis studied the safety of the applied power
For lbt = 20 mm and Ghw bt = 1.64 dB, Gp = 8 dB, and the sensitivity to the antenna-head probe distance. For
brn rx brn
αphn = 1 dB/cm and rbt = 100 mm at fphn = 1 GHz, the safety for the applied input power of PEGRH = 10dBm,

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in IEEE Sensors Journal. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/JSEN.2023.3341168
AUTHOR et al.: PREPARATION OF PAPERS FOR IEEE TRANSACTIONS AND JOURNALS (FEBRUARY 2017) 5

corresponding measured matrix is transformed to the time

domain and delayed by n × ∆t (where n=1,2..50), resulting
in a PAM received signal, which contains the UWB pulses
enveloped by the AP, as presented for 1ms period of the
simulated signal in Fig.6(a), and in Fig.6(b) for the measured
one, where we can observe the modulation of the UWB pulses
with the functional signal modeling the brain activity.

Fig. 4. Specific Absorption Rate (SAR) for the human head model for
1g and 10g of tissue.

the Specific Absorption Rate (SAR) is shown in Fig.4 for

1g and 10g of human tissue, where the obtained value at
1GHz is 0.62W/kg and 0.20W/kg, respectively, are below
the limit (1.60W/kg and 2.00W/kg). For an antenna-head (a)
tissue separation sensitivity, it was seen that changes up to
10mm do not significantly affect either the dimension or
the location of the reconstructed PD region. In Fig.5, the
experimental set-up used for the functional monitoring is
presented, where the VNA is connected to two PC controlled
RF-switches that create the transmitting and receiving links
from the VNA’s ports to the probe antennas.

(b)

Fig. 6. The backscatter pulse amplitude modulated signal. (a) The

numerical results. (b) The experimental results.

After demodulating the signals for a time extension of

10 ms (10 periods of the signal in Fig.6), the AP obtained
is presented in Fig.7, where we can identify and compare the
functional signal obtained by the bio-tag located within the
head-brain model for the simulated and measured setup. An

Fig. 5. The experimental hardware diagram.

The results were extracted during a time range of Tmax =

10 ms, in which the Matlab script is used to collect the data
from the CST software and from the VNA for the numerical
and the experimental validation processes, respectively, as
previously mentioned in Fig.1.
A signal generator is used to create the AP shape on the bio-
tag within the small signals’ regime, modulating the UWB RF
signals transmitted by the probes using the Pulse Amplitude
Modulation (PAM).
The UWB scattering parameters are extracted for 50 sam-
ples, with a time step between samples of ∆t = 20 µs. The Fig. 7. The numerical and the experimental extracted AP signal.

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in IEEE Sensors Journal. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/JSEN.2023.3341168
6 IEEE SENSORS JOURNAL, VOL. XX, NO. XX, XXXX 2017

AP firing rate of fbt = 1 kHz may be observed in accordance model are presented in different colors to precisely localize
to the modulating signal produced by the active zone of the the bio-tag within the brain, which is shown in the right part
photodiode, mimicking the real brain spiking. of the basal ganglia. The experimental reconstructed image
The subtraction of the scattered parameter measurement locates precisely the origin of the bio-tag used to generate the
for the two ON and OFF states will allow keeping the brain activity within the zone of interest, as in Fig.8(d).
mod
usefully modulated component and to remove the non-desired In Table II, the comparison of the differential ∆S21
reflections coming from the measurement environment, re- values for the analytical, numerical, and experimental cases is
sulting on the signal represented in Fig.7. Then by applying presented. The results show a good agreement and also prove
the Multi-Frequency Bi-Focusing (MFBF) algorithm [51], we the capability of locating and extracting the low-level signals
reconstruct the microwave image that allows locating the bio- from a high permittivity medium as the human body. These
tag (modeling the brain activity within the PD region) as in can help in monitoring the functional diseases, as presented in
Fig.8 for the numerical and experimental results. the next section.
In Fig.8(a), a realistic human head model used for the nu- TABLE II
merical validation with skin, skull, CSF, grey, and white matter D IFFERENTIAL M ODULATED FIELD OF THE PAM
is presented, where the red circle refers to the studied region Analytical differential Numerical differential Experimental differential
related to the PD. The bio-tag is used to model the functional field field field
signals produced on the right side of the basal ganglia to -92 dB -94 dB -95 dB
approach the near-to-reality scenario. The model is validated
numerically with CST software using the system mentioned in An additional study for the effect of the changes in dielectric
Section II, as in Fig.8(b). In Fig.8(c), the numerical microwave properties on the capability of the system to properly locate
image reconstructed for the realistic whole and parts head the origin of the signals was proposed, where a 10% change

(a) (b)

(c) (d)

Fig. 8. The reconstructed numerical and experimental microwave images. (a) The basal ganglia is illustrated in a realistic human brain. (b) The
numerical simulation of the human head voxel model in CST. (c) The numerical reconstructed image of the human voxel model. (d) The experimental
reconstructed image of the human head phantom.

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in IEEE Sensors Journal. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/JSEN.2023.3341168
AUTHOR et al.: PREPARATION OF PAPERS FOR IEEE TRANSACTIONS AND JOURNALS (FEBRUARY 2017) 7

in permittivity (εr = 57 ± 5.7) results in a 5% location error,

which is not significant for the purpose of locating the region
where the brain activity related to the Parkinson’s disease

IV. E XPERIMENTAL VALIDATION OF THE BETA

OSCILLATION CHANGES FOR PD

The proposed technique is now explored to study its

capability to monitor the changes in the beta frequency
band oscillatory character that happens in the basal gan-
glia–thalamocortical (BGTC) as previously mentioned [14],
[30]–[32]. To evaluate this part, we propose a beta signal
inspired from [14] to see the capability of the system on Fig. 10. PD And Normal States of the extracted Beta Brain activity
model.
differentiating between the two signals (healthy and unhealthy)
in terms of shape, and duration (firing rate). The PD’s signal
is modeled with a higher value dual peak and longer duration V. C ONCLUSION
(lower firing rate) compared to the normal one.
The experimental analysis of differentiating the PD’s state This paper has presented a novel methodology for monitor-
from the healthy one is conducted using the same setup, the ing functional neural diseases, and in particular, the Parkinson
corresponding functional activity signals were produced in the disease (PD). To model the functional activity within the brain
bio-tag and the scattering signals were measured for 150 ms. medium, a bio-tag has been designed and tested on generating
The received PAM signals are presented in Fig.9(a) for the the AP signal and the proposed technique has proven to be
normal state, and in Fig9(b) for the PD’s state. able to extract it from the received PAM signals, as well as
By analyzing the two signals in Fig.10, it is observed a locating the origin of those activities using UWB microwave
frequency difference of 25% (15Hz for the PD state and 20Hz imaging. Based on this technique, we modeled the normal
for the normal state). This frequency difference corresponds and Parkinson beta oscillation basal ganglia–thalamocortical
with the values in [52], as mentioned in the introductory part, (which is one of the PD markers), and as a combined result
based on both the firing rate and the shape of the signal [53] we have been able to extract and locate its origin precisely.
This technique will help on monitoring the brain to detect PD
based on the firing rate and brain signal shape.

ACKNOWLEDGEMENTS
This work was financially supported in part by CICYT
PID2019- 107885GB-C31, PID2022-136869NB-C31, 2021
SGR 01415, Metropolis PLEC2021-007609, (FI-SDUR)(2021
FISDU 00195), PDR-2014-2022/56-30157-2021-2 A grant
of the Generalitat de Catalunya, and Prueba de Concepto
PDC2022-133091-I00 grants of the Ministerio de Ciencia,
Innovación y Universidades (Spain).

(a)
R EFERENCES
[1] Y. Akazzim, O. El Mrabet, J. Romeu, and L. Jofre-Roca, “Multi-element
uwb probe optimization for medical microwave imaging,” Sensors,
vol. 23, no. 1, p. 271, 2022.
[2] Y. Akazzim, M. Jofre, O. El Mrabet, J. Romeu, and L. Jofre-Roca,
“Uwb-modulated microwave imaging for human brain functional mon-
itoring,” Sensors, vol. 23, no. 9, p. 4374, 2023.
[3] C. Gilmore, A. Zakaria, S. Pistorius, and J. LoVetri, “Microwave
imaging of human forearms: Pilot study and image enhancement,”
Journal of Biomedical Imaging, vol. 2013, pp. 19–19, 2013.
[4] S. Alani, Z. Zakaria, T. Saeidi, A. Ahmad, M. A. Imran, and Q. H.
Abbasi, “Microwave imaging of breast skin utilizing elliptical uwb
antenna and reverse problems algorithm,” Micromachines, vol. 12, no. 6,
p. 647, 2021.
[5] M. Bashri, T. Arslan, W. Zhou, and N. Haridas, “A compact rf switch-
(b) ing system for wearable microwave imaging,” in 2016 Loughborough
Antennas & Propagation Conference (LAPC). IEEE, 2016, pp. 1–4.
Fig. 9. The measured backscattered PAM signals corresponding to the [6] O. Babarinde, M. Jamlos, P. Soh, D.-P. Schreurs, and A. Beyer, “Mi-
brain beta band oscillatory activity modeling. (a) The normal state. (b) crowave imaging technique for lung tumour detection,” in 2016 German
The PD state. Microwave Conference (GeMiC), 2016, pp. 100–103.

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in IEEE Sensors Journal. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/JSEN.2023.3341168
8 IEEE SENSORS JOURNAL, VOL. XX, NO. XX, XXXX 2017

[7] M. Islam, M. Mahmud, M. T. Islam, S. Kibria, and M. Samsuzzaman, in midbrain neurons by multifunctional diamond-based microarrays,”
“A low cost and portable microwave imaging system for breast tumor Frontiers in neuroscience, vol. 13, p. 288, 2019.
detection using uwb directional antenna array,” Scientific reports, vol. 9, [30] T. Wichmann, H. Bergamn, and M. DeLong, “Changes in motor behavior
no. 1, p. 15491, 2019. and neuronal activity in the internal pallidum induced by subthalamic
[8] M. E. Gharbi, M. Martinez-Estrada, R. Fernández-Garcı́a, and I. Gil, inactivation in the mptp model of parkinsonism,” J Neurophysiol, vol. 72,
“Determination of salinity and sugar concentration by means of a pp. 521–30, 1994.
circular-ring monopole textile antenna-based sensor,” IEEE Sensors [31] A. Nini, A. Feingold, H. Slovin, and H. Bergman, “Neurons in the globus
Journal, vol. 21, no. 21, pp. 23 751–23 760, 2021. pallidus do not show correlated activity in the normal monkey, but phase-
[9] S. R. Mohd Shah, N. B. Asan, J. Velander, J. Ebrahimizadeh, M. D. locked oscillations appear in the mptp model of parkinsonism,” Journal
Perez, V. Mattsson, T. Blokhuis, and R. Augustine, “Analysis of thick- of neurophysiology, vol. 74, no. 4, pp. 1800–1805, 1995.
ness variation in biological tissues using microwave sensors for health [32] A. Devergnas, D. Pittard, D. Bliwise, and T. Wichmann, “Relation-
monitoring applications,” IEEE Access, vol. 7, pp. 156 033–156 043, ship between oscillatory activity in the cortico-basal ganglia network
2019. and parkinsonism in mptp-treated monkeys,” Neurobiology of disease,
[10] A. Cataldo, E. De Benedetto, R. Schiavoni, G. Monti, A. Tedesco, vol. 68, pp. 156–166, 2014.
A. Masciullo, E. Piuzzi, and L. Tarricone, “Portable microwave reflec- [33] Mohammed, I. M. Khan, S. Khan, J. Chebil, KhalidA.S.Al, and Khateeb,
tometry system for skin sensing,” IEEE Transactions on Instrumentation “Uwb pulse generation and modulation for signal extraction from
and Measurement, vol. 71, pp. 1–8, 2022. implantable devices,” 2012.
[11] X. Li, Q. Xia, D. Qu, T. Wu, D. Yang, W. Hao, X. Jiang, and X. Li, “The [34] Z. Lin and P. Wei, “Pulse amplitude modulation direct sequence ultra
dynamic dielectric at a brain functional site and an em wave approach wideband sharing signal for communication and radar systems,” in 2006
to functional brain imaging,” Scientific reports, vol. 4, no. 1, p. 6893, 7th International Symposium on Antennas, Propagation & EM Theory.
2014. IEEE, 2006, pp. 1–5.
[12] X. Jiang, Z. Geng, X. Li, L. Peng, B. Kang, and C. Zheng, “Microwave [35] A. Ritzau-Jost, I. Delvendahl, A. Rings, N. Byczkowicz, H. Harada,
transmission approach for dynamic dielectric detection at brain func- R. Shigemoto, J. Hirrlinger, J. Eilers, and S. Hallermann, “Ultrafast
tional site,” in 2017 IEEE MTT-S International Microwave Symposium action potentials mediate kilohertz signaling at a central synapse,”
(IMS). IEEE, 2017, pp. 1235–1238. Neuron, vol. 84, no. 1, pp. 152–163, 2014.
[13] J.-K. Wang, X. Jiang, L. Peng, X.-M. Li, H.-J. An, and B.-J. Wen, [36] C. Neudorfer, C. T. Chow, A. Boutet, A. Loh, J. Germann, G. J. Elias,
“Detection of neural activity of brain functional site based on microwave W. D. Hutchison, and A. M. Lozano, “Kilohertz-frequency stimulation
scattering principle,” IEEE Access, vol. 7, pp. 13 468–13 475, 2019. of the nervous system: A review of underlying mechanisms,” Brain
[14] Y. Yu, D. E. Sanabria, J. Wang, C. M. Hendrix, J. Zhang, S. D. Nebeck, stimulation, vol. 14, no. 3, pp. 513–530, 2021.
A. M. Amundson, Z. B. Busby, D. L. Bauer, M. D. Johnson et al., [37] S. Rashid, L. Jofre, A. Garrido, G. Gonzalez, Y. Ding, A. Aguasca,
“Parkinsonism alters beta burst dynamics across the basal ganglia–motor J. O’Callaghan, and J. Romeu, “3-d printed uwb microwave bodyscope
cortical network,” Journal of Neuroscience, vol. 41, no. 10, pp. 2274– for biomedical measurements,” IEEE Antennas and Wireless Propaga-
2286, 2021. tion Letters, vol. 18, no. 4, pp. 626–630, 2019.
[15] I. Saied and T. Arslan, “Wideband textile antenna for monitoring [38] “Recommended circuit responsivity graph,” 2017. [Online]. Available:
neurodegenerative diseases,” in 2018 IEEE 29th Annual International https://www.thorlabs.com/thorproduct.cfm?partnumber=FDS015
Symposium on Personal, Indoor and Mobile Radio Communications
[39] B. J. Mohammed and A. M. Abbosh, “Realistic head phantom to
(PIMRC), 2018, pp. 356–360.
test microwave systems for brain imaging,” Microwave and Optical
[16] K. Wirdefeldt, H.-O. Adami, P. Cole, D. Trichopoulos, and J. Mandel,
Technology Letters, vol. 56, no. 4, pp. 979–982, 2014.
“Epidemiology and etiology of parkinson’s disease: a review of the
[40] J. Sonne, V. Reddy, and M. R. Beato, “Neuroanatomy, substantia nigra,”
evidence,” European journal of epidemiology, vol. 26, pp. 1–58, 2011.
in StatPearls [Internet]. StatPearls Publishing, 2022.
[17] W. H. Organization. Parkinson disease. [Online]. Available: https:
//www.who.int/news-room/fact-sheets/detail/parkinson-disease [41] W. F. Pickard, “Does the resting potential of chara braunii have an
electrogenic component?” Canadian Journal of Botany, vol. 51, no. 4,
[18] P. P. Michel, E. C. Hirsch, and S. Hunot, “Understanding dopaminergic
pp. 715–724, 1973.
cell death pathways in parkinson disease,” Neuron, vol. 90, no. 4, pp.
675–691, 2016. [42] S. H. Wright, “Generation of resting membrane potential,” Advances in
[19] H. Bernheimer, W. Birkmayer, O. Hornykiewicz, K. Jellinger, and physiology education, vol. 28, no. 4, pp. 139–142, 2004.
F. . Seitelberger, “Brain dopamine and the syndromes of parkinson [43] G. Franceschetti and I. Pinto, “Cell membrane nonlinear response to an
and huntington clinical, morphological and neurochemical correlations,” applied electromagnetic field,” IEEE transactions on microwave theory
Journal of the neurological sciences, vol. 20, no. 4, pp. 415–455, 1973. and techniques, vol. 32, no. 7, pp. 653–658, 1984.
[20] K. Marek and D. Jennings, “Can we image premotor parkinson disease?” [44] M. Jofre, L. Jofre, and L. Jofre-Roca, “On the wireless microwave sens-
Neurology, vol. 72, no. 7 Supplement 2, pp. S21–S26, 2009. [Online]. ing of bacterial membrane potential in microfluidic-actuated platforms,”
Available: https://n.neurology.org/content/72/7 Supplement 2/S21 Sensors, vol. 21, no. 10, p. 3420, 2021.
[21] R. Prashanth, S. D. Roy, P. K. Mandal, and S. Ghosh, “High-accuracy [45] “Fiber-coupled led, 455 nm,” 2017. [Online]. Available: https:
detection of early parkinson’s disease through multimodal features and //www.thorlabs.com/thorproduct.cfm?partnumber=M455F3
machine learning,” International journal of medical informatics, vol. 90, [46] Á. C. Aznar, J. R. Robert, J. M. R. Casals, L. J. Roca, S. B. Boris, and
pp. 13–21, 2016. M. F. Bataller, Antenas. Univ. Politèc. de Catalunya, 2004.
[22] K. R. Gandhi and A. Saadabadi, “Levodopa (l-dopa),” in StatPearls [47] P. V. Nikitin, K. Rao, and R. D. Martinez, “Differential rcs of rfid tag,”
[Internet]. StatPearls Publishing, 2022. Electronics letters, vol. 43, no. 8, p. 1, 2007.
[23] P. S. Larson, “Deep brain stimulation for movement disorders,” Neu- [48] J. C. Bolomey, S. Capdevila, L. Jofre, and J. Romeu, “Electromagnetic
rotherapeutics, vol. 11, pp. 465–474, 2014. modeling of rfid-modulated scattering mechanism. application to tag
[24] T. M. Herrington, J. J. Cheng, and E. N. Eskandar, “Mechanisms of performance evaluation,” Proceedings of the IEEE, vol. 98, no. 9, pp.
deep brain stimulation,” Journal of neurophysiology, vol. 115, no. 1, pp. 1555–1569, 2010.
19–38, 2016. [49] D. M. Lovinger, “Communication networks in the brain,” Alcohol
[25] M. Delenclos, D. R. Jones, P. J. McLean, and R. J. Uitti, “Biomarkers in research and.
parkinson’s disease: Advances and strategies,” Parkinsonism & related [50] P. Le Floch, Q. Li, Z. Lin, S. Zhao, R. Liu, K. Tasnim, H. Jiang, and
disorders, vol. 22, pp. S106–S110, 2016. J. Liu, “Stretchable mesh nanoelectronics for 3d single-cell chronic elec-
[26] X.-X. Wang, Y. Feng, X. Li, X.-Y. Zhu, D. Truong, W. G. Ondo, and trophysiology from developing brain organoids,” Advanced Materials,
Y.-C. Wu, “Prodromal markers of parkinson’s disease in patients with vol. 34, no. 11, p. 2106829, 2022.
essential tremor,” Frontiers in Neurology, vol. 11, p. 874, 2020. [51] L. Jofre, A. Broquetas, J. Romeu, S. Blanch, A. P. Toda, X. Fabregas,
[27] E. Hustad and J. O. Aasly, “Clinical and imaging markers of prodromal and A. Cardama, “Uwb tomographic radar imaging of penetrable and
parkinson’s disease,” Frontiers in neurology, vol. 11, p. 395, 2020. impenetrable objects,” Proceedings of the IEEE, vol. 97, no. 2, pp. 451–
[28] Y. Wu, W. Le, and J. Jankovic, “Preclinical biomarkers of parkinson 464, 2009.
disease,” Archives of neurology, vol. 68, no. 1, pp. 22–30, 2011. [52] L. Iskhakova, P. Rappel, M. Deffains, G. Fonar, O. Marmor, R. Paz,
[29] G. Tomagra, F. Picollo, A. Battiato, B. Picconi, S. De Marchis, Z. Israel, R. Eitan, and H. Bergman, “Modulation of dopamine tone
A. Pasquarelli, P. Olivero, A. Marcantoni, P. Calabresi, E. Carbone induces frequency shifts in cortico-basal ganglia beta oscillations,”
et al., “Quantal release of dopamine and action potential firing detected Nature communications, vol. 12, no. 1, p. 7026, 2021.

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in IEEE Sensors Journal. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/JSEN.2023.3341168
AUTHOR et al.: PREPARATION OF PAPERS FOR IEEE TRANSACTIONS AND JOURNALS (FEBRUARY 2017) 9

[53] C. Liu, G. Zhao, Z. Meng, C. Zhou, X. Zhu, W. Zhang, J. Wang, H. Li, Otman El Mrabet Received the M.Sc. & Ph.D.
H. Wu, C. Fietkiewicz et al., “Closing the loop of dbs using the beta degrees from the University of Abdelmalek Es-
oscillations in cortex,” Cognitive Neurodynamics, vol. 15, pp. 1157– saadi, Morocco, in 2000 and 2004, respectively.
1167, 2021. From March to October 2005, he was with the
Rennes Institute of Electronics and Telecom-
munications, France, as a Visiting Researcher.
From 2007 to 2009, he was a Post-Doctoral
Researcher, under the AECI Fellowship, with
the Electrical and Electronic Engineering De-
partment, UPNA. Since 2009, he has been an
Assistant Professor with the LaSiT laboratory, in
UAE, where he supervises several Ph.D. and M.Sc. theses. a Fulbright
scholar at the University of Wisconsin Madison in 2019. His research in-
terests are UWB antenna design, Metamaterials & Metasurfaces, RFID
tag antennas, rectennas and wireless power transfer, and modeling
active circuits using the finite difference time domain method (FDTD).

Youness akazzim received the License degree Jordi Romeu Ph.D., (IEEE Fellow) He received
in electronics and the M.Sc. degree in Telecom- the Ingeniero de Telecomunicación and Doctor
munication Systems Engineering from Abdel- Ingeniero de Telecomunicación degrees from
malek Essaâdi University, Morocco, in 2016 the Technical University of Catalonia (UPC) -
and 2018, respectively. He is currently pursu- BarcelonaTech (Spain) in 1986 and 1991, re-
ing the Ph.D. degree with the Signal Theory spectively. In 1985, he joined the AntennaLab,
and Communications (TSC) Department, within Signal Theory and Communications Department
the research group of CommSensLab, Universi- (UPC), where he is currently a Full Professor
tat Politècnica de Catalunya (UPC), Barcelona, involved in antennas near-field measurements,
Spain. He held a position as a Research Support electromagnetic scattering and imaging, and
Technician at the Signal Theory and Communi- system miniaturization for wireless and sensing
cations (TSC) Department, UPC, from 2020 to 2021. In 2017, he joined industrial and bio applications. In 1999, he was a Visiting Scholar at
the Information and Telecommunication Systems Laboratory (LaSIT), the Antenna Laboratory of the University of California at Los Angeles
Faculty of Sciences, Tetouan, Morocco, where he has been a Research (USA) on a NATO Scientific Program Scholarship, and at the University
Assistant. He has been awarded the Erasmus + Mobility Grant. He is of California at Irvine (USA) in 2004. He holds several patents, and
currently working on HUMAN BODY PARTS FUNCTIONALITY MONI- has published 60 refereed articles in international journals and 80
TORING USING MICROWAVE IMAGING. conference proceedings. He was a Grand Winner of the European IT
Prize, awarded by the European Commission for his contributions in the
development of fractal antennas in 1998. He has been involved in the
creation of several spin-off companies.
César Palacios Arias Ph. D. Candidate, (IEEE
Student member, 2018). He received the B.S.
degree in Telecommunication and Electronics
Engineering in 2013 from the Private Technical Luis Jofre-Roca Ph.D., (IEEE Fellow, 2010)
University of Loja UTPL (Loja, Ecuador), and He received the M.Sc. (Ing.) and Ph.D.
the M.S. degree in Electronics Engineering from (Doctor Eng.) degrees in Electronic Engi-
the University of Calabria (Cosenza, Italy) in neering (Telecommunication Engineering) from
2017. He is currently pursuing the Ph.D. de- the Technical University of Catalonia (UPC),
gree in the Signal Theory and Communications Barcelona, Spain, in 1978 and 1982, respec-
(TSC) Department, within the research group tively. He has been visiting professor at the École
of microwave interaction with living organisms, Supérieure d’Electricité Paris (1981-82), with the
CommSensLab, UPC. He has held positions at ALCATEL-LUCENT Georgia Institute of Technology, Atlanta (Ful-
(2013-15), Corporación Nacional de Telecomunicaciones-CNT (2015- bright Scholar, 1986-87), and at the University of
15), external researcher at National University of Chimborazo (2018-), California, Irvine, CA (2001-02). Director (1989-
and research support technician at the Signal Theory and Communi- 94) of the Telecommunication Engineering School, UPC, Vicepresident
cations (TSC) Department, UPC (2020-21). He is currently working on of the UPC (1994-2000), and General Director and Secretary for Catalan
micro-systems design and manufacturing for communication with living Universities and Research (2011-2016). Director of the Catalan Re-
organisms and sensing at X-wave frequencies. search Foundation (2002-04), Director of the UPC-Telefonica Chair on
Information Society Future Trends (2003-), Principal Investigator of the
2008-13 Spanish Terahertz Sensing Lab Consolider Project, Director of
the UPC-SEAT Chair on the Future of Automotive, Research Leader
Marc Jofre Ph.D. He obtained his M.Sc. Electri- of the 2017–2020 CommSensLab Maria de Maeztu Project, Academic
cal Engineering (Telecom) degree in 2008 from Director of the European Consortium for Future Urban Mobility (Carnet)
the Technical University of Catalonia (UPC) - and Chairman of the EIT-Urban Mobility European Association. He
BarcelonaTech (Spain), jointly with the Tech- has authored more than 200 scientific and technical papers, reports,
nical University of Delft (The Netherlands). In and chapters in specialized volumes. Research interests include anten-
2013, he obtained his Ph.D. in Photonic Sci- nas, electromagnetic scattering and imaging, system miniaturization for
ences from ICFO - The Institute of Photonic wireless and sensing for industrial, scientific and medical applications.
Sciences (Spain). From 2018, he has held dif- Current work focuses on Connected Reconfigurable Autonomous Ve-
ferent positions at Technical University of Cat- hicles for Urban Mobility, Massive MIMO Antennas and Microorganism
alonia (UPC), FPHAG – Fundació Privada Hos- wireless interaction.
pital Asil de Granollers and Max-Planck Institute
for Quantum Optics. He has published several peer-reviewed articles
and granted/transferred patent applications, and has participated in
several funded projects (national and international; competitive and
non-competitive). He has extensive experience in innovation, research,
project management, intellectual property and exploitation of quantum
technologies for communication and sensing systems, physical charac-
terization of nonlinear materials, biophysical detectors, platforms for the
quantification of microorganisms and health technology.

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/