7310 IEEE SENSORS JOURNAL, VOL. 20, NO.

13, JULY 1, 2020

Early-Stage Detection of Melanoma Skin Cancer

Using Contactless Millimeter-Wave Sensors
Homa Arab , Lydia Chioukh, Student Member, IEEE, Mansoor Dashti Ardakani,
Steven Dufour , Member, IEEE, and Serioja Ovidiu Tatu , Senior Member, IEEE

Abstract —Studies show that the early (stage I) diagno-

sis of cancer increases the survival rate of patients, and
significantly reduces treatment costs. Millimeter-wave radars
can detect skin tumors based on the difference between the
electrical properties of melanomas and healthy skin tissues.
This paper proposes a low-power multitone continuous-wave
radar design at 77 GHz for biomedical imaging applications,
to be used for skin cancer detection. The sensor implemen-
tation is based on a low-cost Miniature Hybrid Microwave
Integrated Circuit (MHMIC) design. The entire sensor was inte-
grated on a 0.127 mm thin ceramic substrate. Two 16-element
patch antenna arrays are implemented in the transmitter
and receiver modules. The receiver is a six-port homodyne
quadrature down-converter. The integrated power detectors are fabricated using HSCH-9161 Schottky diodes. The sensor
was encapsulated in a metallic fixture with dimensions 35 by 45 mm. Two cascaded amplifier circuits are designed and
implemented for amplifying baseband voltages at the output of each power detector. The proposed radar design has the
ability to detect the dielectic properties of tissues, making it suitable for the detection of melanomas with an accuracy on
the order of tens of microns.
Index Terms — Homodyne, six-port, SFCW radar, millimeter-wave, in-phase/quadrature demodulator, reflectometer,
sensor, melanoma skin cancer.

I. I NTRODUCTION by their partner. Some are detected by physicians. Current

diagnosis techniques for living tissues include dermoscopy,
M ELANOMA, the most life-threatening type of skin
cancer, affected more than 90,000 people and caused
the death of more than 9,000 cancer patients in the U.S.
total body photography, macroscopic low-resolution sequential
imaging, X-ray imaging, and magnetic resonance imaging
in 2018 [1]. Early-stage detection, before the cancer spreads (MRI). At the microscopic level, when a higher resolution is
beyond the epidermis, is the key to a full recovery [2]. necessary, biopsy samples are used to determine the type and
The five-year survival rate is 99% for a localized early- stage of the skin cancer [7]–[11].
stage malignant skin tumor, but it can be as low as 20% Various radar systems have been developed for medical
if the tumor spreads to other parts of the body [1], [3]. applications [12]–[20] such as for breast cancer detection [21],
The majority of skin cancers are detected by the patients or [22], for accurate respiration measurements [23], [24], blood
pressure estimation [25]–[27], and for glucose levels detec-
Manuscript received October 28, 2019; revised December 30, 2019; tion [28], [29]. Mm-wave reflectometry has been used with a
accepted January 7, 2020. Date of publication January 27, 2020; set of two non-invasive in-package reflectometers operating at
date of current version June 4, 2020. The associate editor coordi- 42 GHz and 70 GHz for the diagnosis of human tissue diseases
nating the review of this article and approving it for publication was
Dr. Ferran Reverter. (Corresponding author: Homa Arab.) such as early-stage skin cancers [30]. This study describes the
Homa Arab was with the Centre Énergie Matériaux Télécommunica- design of the first mm-wave multitone radar sensor operating
tions, Institut national de la recherche scientifique (INRS), Montreal, QC at 77 GHz for skin cancer detection. The proposed sensor,
H5A 1K6, Canada. She is now with the École Polytechnique de Montréal,
Montreal, QC H3T 1J4, Canada (e-mail: [email protected]). in combination with the ABCDE (Asymmetry, Border, Color,
Lydia Chioukh was with the École Polytechnique de Montréal, Montreal, Diameter, and Evolving size, shape or color) method can be
QC H3T 1J4, Canada. She is now with the Department of Electrical used as an accurate method for the early detection of skin
Engineering, École de Technologie Supérieure, Montreal, QC H3C1K3,
Canada.
cancer. Limited penetration and higher resolution are two of
Mansoor Dashti Ardakani and Serioja Ovidiu Tatu are with the Centre the reasons that lead us to use mm-waves for skin cancer
Énergie Matériaux Télécommunications, Institut national de la recherche diagnosis.
scientifique (INRS), Montreal, QC H5A 1K6, Canada.
Mm-wave radars detect skin cancer based on the differ-
Steven Dufour is with the École Polytechnique de Montréal, Montreal,
QC H3T 1J4, Canada. ence between the electrical properties of melanomas and
Digital Object Identifier 10.1109/JSEN.2020.2969414 normal skin tissues. The permittivity of the human skin

1558-1748 © 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on September 02,2021 at 08:00:38 UTC from IEEE Xplore. Restrictions apply.
 ARAB et al.: EARLY-STAGE DETECTION OF MELANOMA SKIN CANCER USING CONTACTLESS MILLIMETER-WAVE SENSORS 7311

at mm-wave frequencies, which is known to be complex,

is addressed using theoretical modeling and experimental
measurements [31]–[33]. Various models have been proposed
to describe the permittivity of biological tissues as a func-
tion of the frequency, such as the Davidson-Cole theory,
the constant-phase-angle and Raicu theory, and the Maxwell-
Wagner theory. But they are generally only applicable well
below the mm-wave band [34]. Gabriel’s model is based on the
four terms of the Cole–Cole expression used to theoretically
calculate the permittivity of human skin. It was observed by
researchers who analyzed human tissues that cancer tissues
are characterized by significantly higher water content when
compared to normal skin. Water content can vary from 81.7%
Fig. 1. Permittivity ε and electrical conductivity σ of dry skin, wet skin,
to 60.9% from an unhealthy to a healthy skin tissue. It is also melanoma and water.
known that the energy absorption of electromagnetic radiations
in the millimeter wavelength is significantly more important in
TABLE I
tissues with higher water content [35]. It is therefore possible D IELECTRIC P ROPERTIES OF H UMAN S KIN T ISSUES B ASED ON THE
to use the difference of phase, the frequency and variations of F OUR -T ERM C OLE -C OLE E XPRESSION AT 77 GHz
the relative amplitude of the transmitted and reflected signals
from the target to differentiate a melanoma from a healthy skin
tissue. To the best of our knowledge, the proposed radar sensor
front-end is the first fully integrated 77 GHz prototype using
Miniature Hybrid Microwave Integrated Circuit (MHMIC)
technology. We will show that by using the proposed large
bandwidth six-port receiver based sensor, we can detect a
melanoma with size on the order of a millimeter [36]–[39].
The remainder of the paper is organized as follows:
Section II-A describes skin tissue properties in the mm-wave
frequency. Section II-B describes the proposed system theory
and architecture. Section III-A gives the schematic of the cir-
cuit and simulation results using the ADS simulator. A descrip-
tion of the fabricated circuit, and experimental measurement ε in the vacuum is ε0 = 8.85 × 10−12 F/m. These parameters
results, are reported in Sections III-B and III-C. Concluding give the wave propagation velocity
remarks are given in Section IV.
  − 1
με  σ 2
 2
ν = ω/β = 2 1+ ωε +1 . (2)
II. T HEORY AND B ASIC E QUATIONS
A. Skin Tissue Properties in the Microwave
To show how dielectric properties vary with respect to
Wavelength Range
the frequency and the type of target, the permittivity and
The first step for designing a fully operational radar consists conductivity of dry skin, wet skin, melanoma, and water, in the
in studying how the permittivity of various targets influences bandwidth range of 70 − 82 GHz, are illustrated in Fig. 1.
the reflected signals. The main factors that influence a reflected The main dielectric parameters associated with dry skin,
signal are the permittivity ε and the electrical conductivity σ , wet skin, and melanoma are given in TABLE I. They are
whereas the relative magnetic permeability is generally negli- calculated using the four-term Cole-Cole expression. It can be
gible. The relative electrical permittivity εr = ε/ε0 and con- seen that the use of high frequencies limits the penetration of
ductivity of normal skin are considered as reference quantities. the energy of the radar. Low penetration is the main drawback
The propagation constant of a material is given by γ = α+ jβ, associated with mm-wave radar sensor based technologies for
where α is the radar energy attenuation and β is the phase biomedical applications. For skin cancer however, the lack
constant which are calculated in the MKS system of units of deep penetration is not an issue. A radar sensor that can
based on the theoretical formulas: measure the various dielectric properties of skin tissues with

√  σ  1
2
better accuracy is more important in this case.
α = ω εμ 12 1 + ωε −1 ; Radar signals that reach the receiving antenna are modified
 by the target, which is responsible for the attenuation and the
√  σ  1
β = ω εμ 12 1 + ωε +1 2, (1) absorption of the transmitted signal. This is taken into account
by the model through the determination of the electrical
where the dielectric permeability μ in the vacuum is given by conductivity and the relative dielectric permittivity of the
μ0 = 4π × 10−7 H/m, the angular velocity ω = 2π f rad/s target. The reflected and transmitted signals from a dielec-
where f is the frequency in Hz, and the dielectric permittivity tric interface, S11 and S21 respectively, can be theoretically

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on September 02,2021 at 08:00:38 UTC from IEEE Xplore. Restrictions apply.
 7312 IEEE SENSORS JOURNAL, VOL. 20, NO. 13, JULY 1, 2020

where ARX is the amplitude of the received signal, R is the

range of the target, λ is the wavelength of the carrier signal,
and c is the speed of light. The attenuation in amplitude of
the received signal depends on the dielectric properties of the
targets. As can be seen in Fig. 2, the transmitted signal is used
as a local oscillator (LO) signal to down-convert the received
signal as
 
SBX (t) = ABX cos 4πλR + θ (t) − θ (t − 2R
c ) . (6)
The residual phase noise, θ (t) − θ (t − 2R
c ), is negligible due
to the coherent nature of the sensor. In fact, both the TX and
the LO signals have the same source and phase noise which
Fig. 2. Schematic of the designed radar system in ADS. can cancel each other, as can be seen in the case of short
range sensor applications [21]. The difference between the
received signal and the transmitted waveform can basically
described as:
√ √ be characterized using three components: the amplitude, the
εr − ε0 frequency, and the time delay. In this project, since the targets
S11 = − √ √ ;
εr + ε0 are stationary, they produce various reflected signals with
S21 = (1 − S11
2
) exp (ks ts ), (3) different amplitudes and phases which are related to the range
√ of the targets.
where ks = k0 εr is the free space propagation constant
(k0 = 8.1632 × 10−11 at 77 GHz), and ts is the thickness of III. P ROTOTYPE M EASUREMENTS AND S IMULATIONS
the sample. For example, the thickness of the skin varies from A. Simulation Set-Up and Results
0.5 mm on the eyelids to 4.0 mm on the heels of the feet.
To validate the proposed design, various parts of the cir-
The analysis assumes that the incident signal is a continuous
cuit are modeled using the Advanced Design System (ADS)
wave and that it is normally incident on the dielectric surface
software simulator of Keysight Technologies. The ratio of
which consists in the top layer of the skin. It was already
the received power PRX to the transmitted power PTX can be
mentioned that high mm-wave frequencies do not allow for
calculated using the radar equation,
a deep penetration of energy. The deeper skin layers are
 λ 2  4πσ   λ 2
therefore not taken into consideration in the model. PRX = PTX G TX G RX RCS
, (7)
4π R λ2 4π R
B. Proposed CW Radar Theory where G RX is the receiver gain, G TX is the transmitter gain, and
A radar system is composed of a transmitter that emits an σRCS represents the radar cross section (RCS). Two LSO (Line-
electromagnetic signal, modulated or unmodulated, with a par- Of-Sight) antenna links are used to model the Friis equation in
ticular type of waveform which depends on the requirements of ADS, which includes the modeling of the TX and RX antenna
the application. The signal is reflected back from the target and gains and the modeling of the path losses.
detected by the radar receiver, and is then analyzed in order to The ADS block diagram of the proposed radar system
extract the desired information. The main purpose of a receiver is illustrated in Fig. 2. As a target, an amplifier is used
is to down-convert a received high-frequency signal, used for with specific transmission and reflection coefficients, based on
the transmission, to the baseband signal in order to extract equations (4) whand
2 (5). For a short-range sensor, we consider
its information. This frequency conversion step is usually σRCS = 4π λ as the near-field RCS of the target [40],
performed using various types of mixers. In this research, an where w and h are the width and the height of the target,
innovative measurement technique is used, based on the six- respectively. Using this expression for modeling the target gave
port receiver principle. The proposed six-port circuit offers an us better results with the ADS simulator, when compared to
alternative to the down-converter receivers used today, and it theoretical and measurement results. The path loss is high due
can perform better for some applications. The highly accurate to the millimeter range of the wavelength. Also, the received
phase resolution offered by this alternate technology translates power signal is reduced by 12 dB as the range between the
into better measurement capabilities. The block diagram of the sensor and the target doubles. In order to have a strong enough
proposed sensor system is illustrated in Fig. 2. signal at the baseband output, two high-gain amplifiers are
The signal transmitted by the generator is given by used at the outputs of the radar sensor. The millimeter source
  frequency is set to 77 GHz, and the input power is 0 dBm.
STX (t) = ATX cos 2π f t + θ (t) , (4) The TX and RX antenna gains are 16 dBi and the target range
is 10 cm.
where f is the frequency of the transmitted signal, θ (t) is the
Fig. 3 illustrates plots of the reflection and transmission
phase noise from the RF generator, and ATX is the amplitude
coefficients as a function of the frequency. These simulation
of the transmitted signal. The received signal can then be
results are in a good agreement with the analytical results.
approximated by
  We observe that the reflection and transmission coefficients
SRX (t) = ARX cos 2π f t − 4πλR + θ (t − 2R
c ) , (5) decrease slightly as the frequency increases. The computed I

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on September 02,2021 at 08:00:38 UTC from IEEE Xplore. Restrictions apply.
 ARAB et al.: EARLY-STAGE DETECTION OF MELANOMA SKIN CANCER USING CONTACTLESS MILLIMETER-WAVE SENSORS 7313

TABLE II
D ESIGNED R ESISTORS U SED FOR THE F INAL
B ASEBAND C IRCUIT D ESIGN

as the designed six-port circuit. The simulated peak gains are

6.3 dBi for the element and 16.2 dBi for the array at 77 GHz.
The four output ports of the six-port circuit are connected
to power detectors that deliver a voltage that is linearly related
to the power of the RF signal. The output voltage of the four
identical power detectors is proportional to the square of the
magnitude of the RF input signal.
The output impedance of the HSCH-9161 millimeter-wave
Schottky diodes, which are used as power detectors, is in the
M range while the standard impedance for output cables
Fig. 3. Simulation results for the transmission and reflection coefficients.
and measurement instruments is 50 . Moreover, in order
to amplify the received signal and to solve the mismatch
impedance problem, baseband amplifier circuits are designed
and fabricated. The diagram of the baseband circuit, using
current AD8000 feedback operational amplifiers, is illustrated
in Fig. 6. The resistors used are listed in TABLE II. This
very high-speed amplifier has a maximum cut-off frequency
of 1.5 GHz, which is related to its gain. For this reason,
we have designed two modules to achieve an acceptable gain
as well as the desired bandwidth. At each stage of the proposed
design, the six-port DC output voltages are amplified by about
Fig. 4. Simulation results for the four samples under study, for a
frequency range of 1.5 GHz. 19 dB. Also, the non-inverting input impedance of this op-amp
is 2 M while the output impedance of the six-port circuit is
set to 50 . This means that this baseband circuit design is
and Q signals for the four target types are illustrated in Fig. 4 a good candidate for solving the high gain problem, and for
over a bandwidth range of 76.25 – 77.75 GHz. Targets with a converting the high impedance to a desired level.
cross section of 0.01 m2 were located at a distance of 10 cm All the components of the designed radar are fabricated on
from the transmitter and receiver antenna, and frequency the same 0.127 mm ceramic substrate, with the exception of
is sweeped in increments of  f = 60 MHz. As expected, the baseband amplifier which is the only component seperated
simulation points move around a circles with a specific phase from the rest of the circuit. Having all the components of
shift of 14.4◦. the proposed sensor circuit on the same die reduces the
size of the circuit and the number of connections, resulting
in reduced losses due to connections and in a physically
B. Fabricated Circuit
more robust circuit. This robust and low-cost design allows
The proposed radar system, as previously mentioned, for the fabrication of sensors that can be used for various
is based on a six-port interferometer, and an I/Q demodulation biomedical applications. The fabricated circuit based on this
technique is used to recover the desired information. Accord- design was tested for range measurements which resulted in
ing to the theory of six-port reflectometers, the returned signal an accuracy of 20 μm for the whole prototype (measurement
is mixed with the transmitted signal. As illustrated in Fig. 5, based) [41], [42].
the proposed six-port interferometer is a linear passive mm- The WR12 to microstrip transmission line, and the diode-
wave circuit composed of a combination of four 90◦ hybrid based power detector, also have a bandwidth on the order
couplers (HC) with a 90◦ phase shifter, implemented using of 12 GHz. The microstrip patch antenna array reduces the
microstrip transmission lines. The transmitted signal is injected overall radar sensor bandwidth to around 6 GHz, covering the
at port 5 as a local reference signal, and at port 6 as a received available ISM band between 75.5 GHz and 81.5 GHz.
signal. This circuit has been fabricated on a 127 μm thin
ceramic substrate, with a relative permittivity of εr = 9.8,
using in-house MHMIC technology. C. Measurement Set-Up
A 8 × 2 microstrip patch antenna array is used for two Measurements were performed using the fabricated proto-
reasons. First, it is simple to design and fabricate. Second, type, with the experimental set-up illustrated in Fig. 7. It is
it is small and can be fabricated on the same ceramic substrate composed of a HP-83550 series mm-wave source, a mm-wave

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on September 02,2021 at 08:00:38 UTC from IEEE Xplore. Restrictions apply.
 7314 IEEE SENSORS JOURNAL, VOL. 20, NO. 13, JULY 1, 2020

Fig. 5. Block diagram and prototypes of various components of the designed radar sensor.

TABLE III
M EASUREMENT R ESULTS FOR VARIOUS T ISSUES

bandwidth. Four baseband amplifiers are connected to the

Fig. 6. Block diagram of two cascaded baseband operational amplifiers.
output ports of the sensor using cables.
Preliminary I and Q signal measurements were performed
in our laboratory. Targets with a cross section of 0.01 m2 were
located at a distance of 10 cm from the fabricated sensor,
and phase shifts were measured for various frequencies in
increments of  f = 60 MHz from 76.25 GHz to 77.75 GHz.
Due to laboratory related limitations, the measurements were
performed for three different samples: water, a wet hand, and a
dry hand. Experimental results are illustrated in Fig. 8 for each
frequency increment, over a bandwidth of 1.5 GHz. Given the
noise level and the offset error around each measurement point,
it was not possible to use raw measurement data to calculate
the phase difference between each point. Measurement results
are therefore illustrated in the IQ– plane, using an average of
the measurements with the lowest amount of noise.
Fig. 7. Measurement set-up. As expected, measurement points move around ellipses
with a specific phase shift of 14.4◦. This agrees with the
expected target range given by equation (7). The radius of
active multiplier from OML (X6), and a Tektronix DPO each trajectory is calculated using the average of the length
7054 digital phosphor oscilloscope. The millimeter source of the semi-major and of the semi-minor axes of the ellipses.
frequency is set to 12.83 GHz with a 10 dBm input power. The DC offset is subtracted from the data in order to have
The multiplier will then generate a 77 GHz signal with 0 dBm the center of the I/Q trajectory centered at the origin. This
power at the input of the parallel line coupler. The die is does not have any effect on the target range and amplitude
mounted and power detectors are wire-bonded to the PCB measurement results. The radius of each trajectory represents
using ribbon bonds to ensure good matching over a broad the transmission coefficient of each target and the area of each

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on September 02,2021 at 08:00:38 UTC from IEEE Xplore. Restrictions apply.
 ARAB et al.: EARLY-STAGE DETECTION OF MELANOMA SKIN CANCER USING CONTACTLESS MILLIMETER-WAVE SENSORS 7315

Fig. 9. Measurement results for spectrum of the I and Q signals

at 77 GHz.

TABLE IV
Fig. 8. Measurement results for I and Q signals for various samples, C OMPARISON OF B IOMEDICAL M ILLIMETER -WAVE R ADAR S ENSORS .
while the frequency sweeps over 1.5 GHz.

ellipse (a perfect circle in theory) is directly proportional to

the received power.
Fig. 9 shows the spectra of the IF signal for three different
targets at 77 GHz. A flat metal plate is used as a reference
target. Three different target peaks from I and Q signal six-port calibration. The whole circuit dimensions (WR12 to
measurements show the measured attenuation for the various MSTL transition, power detectors, six-port interferometer, and
samples. A slight difference in peak position can be seen TX/RX antenna) is 5.49 × 2.54 cm with low-manufacturing
between I and Q. This can be related to many factors such price when compared to similar prototypes using the same
as the characteristics of the six-port circuit, the differences technology. The six-port-based receiver has a 12 GHz band-
between the four power detectors at the output ports, and the width with an accuracy of 10 μm, which is related to the
settings of the oscilloscope if DC coupling is used. circuit phase errors. The WR12 to microstrip transmission line,
The most relevant publications related to this manuscript are and the diode-based power detector, also have a bandwidth
compared in the Table IV below. They are all based on CW around 12 GHz. The microstrip patch antenna array reduces
radar sensors with very simple waveguides or transmission the overall radar sensor bandwidth to around 6 GHz, covering
line probes, or on antenna prototypes. Unfortunately, many the available ISM band between 75.5 GHz and 81.5 GHz.
of those methods require that the materials used for testing
to be in contact with the sensor or cut in thin slices for IV. C ONCLUSION
measuring transmission and reflection coefficients. In addition, This work shows the feasibility of using mm-wave radars for
these studies are different with respect to center frequencies, biomedical applications. A novel mm-wave integrated multi-
bandwidth, and range resolution. tone radar sensor for melanoma detection is proposed. Dif-
In our study, the design and implementation of a whole ferences in dielectric properties of normal and tumor tissues,
prototype of a contactless 77 GHz FMCW radar sensor is when electromagnetic radiation is applied, results in different
studied. It is the first six-port interferometer based FMCW energy absorption. The proposed radar sensor can accurately
radar at 77 GHz on a 127 μm ceramic substrate. It is also the detect materials with different dielectric properties, making it
first hardware design and implementation of a 77 GHz low-cost a promising technology to be used in biomedical applications.
radar sensor designed to obtain raw data without the need for It is integrated using an MHMIC technology and it is mounted

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on September 02,2021 at 08:00:38 UTC from IEEE Xplore. Restrictions apply.
 7316 IEEE SENSORS JOURNAL, VOL. 20, NO. 13, JULY 1, 2020

on a metallic fixture for laboratory measurements. It consists [19] M. Amin, M. Farhat, and H. Baǧci, “A dynamically reconfigurable
of a wide-band six-port interferometer (12 GHz) and two fano metamaterial through graphene tuning for switching and sensing
applications,” Sci. Rep., vol. 3, no. 1, p. 2105, Dec. 2013.
antenna arrays of 16 patch elements. The measured amplitude [20] A. Och et al., “Accuracy bounds and measurements of a contactless
and phase of both quadrature signals generated by the six-port permittivity sensor for gases using synchronized low-cost mm-wave
interferometer is in agreement with simulation and theoretical frequency modulated continuous wave radar transceivers,” Sensors,
vol. 19, no. 15, p. 3351, 2019.
results, for all the tested tissues, from 76.5 GHz to 77.5 GHz. [21] M. Bassi, M. Caruso, M. S. Khan, A. Bevilacqua, A.-D. Capobianco,
This study also describes the first six-port interferometer based and A. Neviani, “An integrated microwave imaging radar with planar
multi-tone radar at 77 GHz, fabricated on a 127 μm ceramic antennas for breast cancer detection,” IEEE Trans. Microw. Theory
Techn., vol. 61, no. 5, pp. 2108–2118, May 2013.
substrate. It is also the first design and implementation of a [22] A. Martellosio et al., “Dielectric properties characterization from 0.5 to
77 GHz low-cost radar sensor designed to collect raw data 50 GHz of breast cancer tissues,” IEEE Trans. Microw. Theory Techn.,
without needing six-port calibration. This robust and low-cost vol. 65, no. 3, pp. 998–1011, Mar. 2017.
[23] E. Schires, P. Georgiou, and T. S. Lande, “Vital sign monitoring through
design allows for the fabrication of sensors that can be used the back using an UWB impulse radar with body coupled antennas,”
for skin cancer treatment. IEEE Trans. Biomed. Circuits Syst., vol. 12, no. 2, pp. 292–302,
Apr. 2018.
[24] Q. Zhang, Y. Zhou, S. Song, G. Liang, and H. Ni, “Heart rate extraction
R EFERENCES based on near-infrared camera: Towards driver state monitoring,” IEEE
[1] T. Talicia, “Cancer facts & figures 2012. American Cancer Society Access, vol. 6, pp. 33076–33087, 2018.
(ACS),” in American Cancer Society. Atlanta, GA, USA: Taylor & [25] S. I. Ganchev, S. Bakhtiari, and R. Zoughi, “A novel numerical technique
Francis, 2012, p. 66. for dielectric measurement of generally lossy dielectrics,” IEEE Trans.
[2] C. Barata, M. Ruela, M. Francisco, T. Mendonca, and Instrum. Meas., vol. 41, no. 3, pp. 361–365, Jun. 1992.
J. S. Marques, “Two systems for the detection of melanomas in [26] J. Solà et al., “Noninvasive and nonocclusive blood pressure estimation
dermoscopy images using texture and color features,” IEEE Syst. J., via a chest sensor,” IEEE Trans. Biomed. Eng., vol. 60, no. 12,
vol. 8, no. 3, pp. 965–979, Sep. 2014. pp. 3505–3513, Dec. 2013.
[3] L. Torre et al., “Global cancer statistics, 2012,” CA: Cancer J. Clini- [27] Y.-L. Zheng et al., “An armband wearable device for overnight and cuff-
cians, vol. 65, no. 2, pp. 87–108, 2015. less blood pressure measurement,” IEEE Trans. Biomed. Eng., vol. 61,
[4] M. Borlu and M. E. Yuksel, “Development of an image processing no. 7, pp. 2179–2186, Jul. 2014.
system for automatic melanoma diagnosis from dermoscopic images: [28] A. E. Omer, G. Shaker, S. Safavi-Naeini, K. Murray, and R. Hughson,
Preliminary,” Turkish J. Dermatol., vol. 2, pp. 111–115, Dec. 2008. “Glucose levels detection using mm-wave radar,” IEEE Sensors Lett.,
[5] O. Abuzaghleh, B. D. Barkana, and M. Faezipour, “Noninvasive real- vol. 2, no. 3, pp. 1–4, Sep. 2018.
time automated skin lesion analysis system for melanoma early detection [29] S. Saha et al., “A glucose sensing system based on transmission
and prevention,” IEEE J. Transl. Eng. Health Med., vol. 3, 2015, measurements at millimetre waves using micro strip patch antennas,”
Art. no. 4300212. Sci. Rep., vol. 7, no. 1, p. 6855, Dec. 2017.
[6] T. Dwyer et al., “The use of spectrophotometry to estimate melanin den- [30] A. Taeb, S. Gigoyan, and S. Safavi-Naeini, “Millimetre-wave waveguide
sity in Caucasians,” Cancer Epidemiol. Prevention Biomarkers, vol. 7, reflectometers for early detection of skin cancer,” IET Microw., Antennas
no. 3, pp. 203–206, 1998. Propag., vol. 7, no. 14, pp. 1182–1186, Nov. 2013.
[7] J. V. Frangioni, “New technologies for human cancer imaging,” J. Clin. [31] IEEE Standard for Safety Levels With Respect to Human Exposure
Oncol., vol. 26, no. 24, pp. 4012–4021, Aug. 2008. to Radio Frequency Electromagnetic Fields 3 kHz to 300 GHz, IEEE
[8] H. F. Zhang, K. Maslov, G. Stoica, and L. V. Wang, “Functional Standard C95.1, 1999.
photoacoustic microscopy for high-resolution and noninvasive in vivo [32] Y. Feldman et al., “The electromagnetic response of human skin in the
imaging,” Nature Biotechnol., vol. 24, pp. 848–851, Jun. 2006. millimetre and submillimetre wave range,” Phys. Med. Biol., vol. 54,
[9] S. W. Menzies et al., “Impact of dermoscopy and short-term sequential no. 11, pp. 3341–3363, 2009.
digital dermoscopy imaging for the management of pigmented lesions [33] H. Bahramiabarghouei, E. Porter, A. Santorelli, B. Gosselin, M. Popovic,
in primary care: A sequential intervention trial,” Brit. J. Dermatol., and L. A. Rusch, “Flexible 16 antenna array for microwave breast cancer
vol. 161, no. 6, pp. 1270–1277, 2009. detection,” IEEE Trans. Biomed. Eng., vol. 62, no. 10, pp. 2516–2525,
[10] A. B. Kimball et al., “Magnetic resonance imaging detection of occult Oct. 2015.
skin and subcutaneous abnormalities in juvenile dermatomyositis: Impli- [34] A. D. Droitcour, O. Boric-Lubecke, V. M. Lubecke, J. Lin, and
cations for diagnosis and therapy,” Arthritis Rheumatism, vol. 43, no. 8, G. T. A. Kovacs, “Range correlation and I/Q performance benefits
pp. 1866–1873, 2000. in single-chip silicon Doppler radars for noncontact cardiopulmonary
[11] C. A. McBain et al., “X-ray volumetric imaging in image-guided monitoring,” IEEE Trans. Microw. Theory Techn., vol. 52, no. 3,
radiotherapy: The new standard in on-treatment imaging,” Int. J. Radiat. pp. 838–848, Mar. 2004.
Oncol. Biol. Phys., vol. 64, no. 2, pp. 625–634, Feb. 2006. [35] R. Deban, H. Boutayeb, K. Wu, and J. Conan, “Deterministic approach
[12] C. Wu and M. Soleimani, “Frequency difference EIT with localization: for spatial diversity analysis of radar systems using near-field radar cross
A potential medical imaging tool during cancer treatment,” IEEE Access, section of a metallic plate,” IEEE Trans. Antennas Propag., vol. 58,
vol. 7, pp. 21870–21878, 2019. no. 3, pp. 908–916, Mar. 2010.
[13] H. Patel, “Microwave imaging for breast cancer detection using 3D level [36] G. F. Engen, An Introduction to the Description and Evaluation of
set based optimization, FDTD method and method of moments,” Ph.D. Microwave Systems Using Terminal Invariant Parameters. Washington,
dissertation, Mar. 2019. DC, USA: National Bureau of Standards, 1969.
[14] Z. Vafapour, “Polarization-independent perfect optical metamaterial [37] H. Arab, C. Akyel, and S. Tatu, “Wide-band millimetre wave down-
absorber as a glucose sensor in food industry applications,” IEEE Trans. converter based on six-port circuit for radar and sensing applications,”
Nanobiosci., vol. 18, no. 4, pp. 622–627, Oct. 2019. in Proc. 32nd Gen. Assem. Sci. Symp. Int. Union Radio Sci., Montreal,
[15] A. Keshavarz and Z. Vafapour, “Water-based terahertz metamaterial for QC, Canada, Aug. 2017, pp. 1–3.
skin cancer detection application,” IEEE Sensors J., vol. 19, no. 4, [38] H. Arab, S. O. Tatu, and C. Akyel, “Performance analysis for two
pp. 1519–1524, Feb. 2019. different structures of 77 GHz six-port correlator,” in Proc. 17th
[16] A. Keshavarz and Z. Vafapour, “Sensing avian influenza viruses using Int. Symp. Antenna Technol. Appl. Electromagn. (ANTEM), Jul. 2016,
terahertz metamaterial reflector,” IEEE Sensors J., vol. 19, no. 13, pp. 1–2.
pp. 5161–5166, Jul. 2019. [39] H. Arab, S. O. Tatu, and C. Akyel, “Design and characterization of a 77
[17] X. Yan et al., “The terahertz electromagnetically induced transparency- GHz six-port modulator for an automobile radar,” in Proc. IEEE 84th
like metamaterials for sensitive biosensors in the detection of cancer Veh. Technol. Conf. (VTC-Fall), Sep. 2016, pp. 1–5.
cells,” Biosensors Bioelectron., vol. 126, pp. 485–492, Feb. 2019. [40] C. Gu, R. Li, C. Li, and S. B. Jiang, “Doppler radar respiration
[18] Z. Vafapour and H. Ghahraloud, “Semiconductor-based far-infrared measurement for gated lung cancer radiotherapy,” in Proc. IEEE Top.
biosensor by optical control of light propagation using THz metama- Conf. Biomed. Wireless Technol., Netw., Sens. Syst., Phoenix, AZ, USA,
terial,” J. Opt. Soc. Amer. B., vol. 35, no. 5, p. 1192, May 2018. Jan. 2011, pp. 91–94.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on September 02,2021 at 08:00:38 UTC from IEEE Xplore. Restrictions apply.
 ARAB et al.: EARLY-STAGE DETECTION OF MELANOMA SKIN CANCER USING CONTACTLESS MILLIMETER-WAVE SENSORS 7317

[41] H. Arab, S. Dufour, E. Moldovan, C. Akyel, and S. O. Tatu, “Accurate Mansoor Dashti Ardakani received the B.Sc. degree in electrical engi-
and robust CW-LFM radar sensor: Transceiver front-end design and neering from the Shiraz University of Technology, Iran, in 2009, and the
implementation,” IEEE Sensors J., vol. 19, no. 5, pp. 1943–1950, M.Sc. degree in telecommunications engineering from the Iran University
Mar. 2019. of Science and Technology (IUST), Iran, in 2012. He is currently pursuing
[42] H. Arab, S. Dufour, E. Moldovan, C. Akyel, and S. O. Tatu, “A 77- the Ph.D. degree with the Institut National de la Recherche Scientifique
GHz six-port sensor for accurate near-field displacement and Doppler (INRS), Montreal, Canada. He worked with Teyf Ertebat Ltd., Tehran,
measurements,” Sensors, vol. 18, no. 8, P. 2565, Aug. 2018. Iran, as an RF Engineer and the Technical Manager from 2011 to
[43] U. A. Khan, N. Al-Moayed, N. Nguyen, K. A. Korolev, M. N. Afsar, 2016. His current research interests involve microwave and millimeter-
and S. P. Naber, “Broadband dielectric characterization of tumorous wave circuits, systems and wireless transceivers, radars, RF modules,
and nontumorous breast tissues,” IEEE Trans. Microw. Theory Techn., antennas, and microwave measurement techniques.
vol. 55, no. 12, pp. 2887–2893, Dec. 2007.
[44] A. Mirbeik-Sabzevari, S. Li, E. Garay, H.-T. Nguyen, H. Wang, and
N. Tavassolian, “Synthetic ultra-high-resolution millimeter-wave imag-
ing for skin cancer detection,” IEEE Trans. Biomed. Eng., vol. 66, no. 1,
pp. 61–71, Jan. 2019.
[45] S. Heunisch, L. Ohlsson, and L.-E. Wernersson, “Reflection of coher-
ent millimeter-wave wavelets on dispersive materials: A study on
porcine skin,” IEEE Trans. Microw. Theory Techn., vol. 66, no. 4,
pp. 2047–2054, Apr. 2018.
[46] M. J. Horst, M. T. Ghasr, and R. Zoughi, “Design of a compact V-Band
transceiver and antenna for millimeter-wave imaging systems,” IEEE
Trans. Instrum. Meas., vol. 68, no. 11, pp. 4400–4411, Nov. 2019.
Steven Dufour (Member, IEEE) received the B.Sc. degree in mathe-
matics and the M.Sc. degree in applied mathematics from the Université
de Montréal, and the Ph.D. degree in engineering mathematics from the
École Polytechnique de Montréal. After two years as an Assistant Profes-
sor with the Department of Mathematics, University of Wyoming, USA, he
Homa Arab was born in Iran. She received the master’s degree in returned to the Department of Mathematics and Industrial Engineering,
telecommunications engineering from École Polytechnique de Montréal École Polytechnique de Montréal, where he is currently an Associate
and the Ph.D. degree in telecommunications from the Centre Énergie Professor. He was a Visiting Professor with UERJ, Brazil, and KAUST,
Matériaux Télécommunications, Institut national de la recherche scien- Saudi Arabia. He teaches applied mathematics to engineering students.
tifique (INRS), Montreal, Canada. She was with Moshanir Company, His current research interests are the development of new finite-element-
Control and Protection Department from 2005 to 2011. She is currently based numerical methodologies for modeling fluid mechanics problems
a Postdoctoral Fellow with École Polytechnique de Montréal. Her current found in free surface and turbulent flows, modeling electromagnetism
research interests are microwave and millimeter-wave circuit design, phenomena found in supraconductors and the optimization of antennas,
transmitter and receivers, radars, and dielectric measurements. and the study of magnetohydrodynamics.

Lydia Chioukh (Student Member, IEEE) received the French D.E.A.

(M.Sc.) degree in microwave and micro-technology from the University
des Sciences et Technologies de Lille, France, in 2003, the M.Sc.
degree in mechatronics from the University d’Artois, Pas de Calais,
France, in 2006, the M.Sc. degree in biomedical engineering from
École Polytechnique Montréal, Montreal, QC, Canada, in 2009, and the
Ph.D. degree in electronics and microwave engineering in 2015. She
is currently a Postdoctoral Researcher with the École de Technologie
Supérieure, Montreal. Her research interests include the fields of radars,
sensors, and biomedical applications of wireless technology. She is a Serioja Ovidiu Tatu (Senior Member, IEEE) received the B.Sc. degree
member of the IEEE Microwave Theory and Techniques Society (IEEE in radio engineering from the Polytechnic University of Bucharest,
MTT-S), the IEEE WIE Society, and the IEEE Engineering in Medicine Romania, in 1989, and the M.Sc.A. and Ph.D. degrees in electrical engi-
and Biology Society. She was a recipient of the Doctoral Fellowships from neering from the École Polytechnique de Montréal, Canada, in 2001 and
Le Fonds de recherche du Québec—Nature et Technologies (FQRNT) 2004, respectively. He was with National Company of Telecommunica-
in 2012, the Best Poster Award of the Center for Radiofrequency Elec- tions, Rom-Telecom, Bistrita-Nasaud, Romania, as an RF Engineer and
tronics Research of Quebec (CREER) Symposium in 2009, the Rohde the Head of the Telecommunications Laboratory from 1989 to 1993, and
and Schwarz Award of the CREER Symposium in 2013, the Larry K. as a Technical Manager from 1993 to 1997. He is currently a Full Profes-
Wilson Regional Student Activities Award in 2011, the IEEE Montreal sor with the Institut national de la recherche scientifique (INRS)–Centre
Section WIE Service Award in 2012, the IEEE Canada Women in Énergie Matériaux Télécommunications, Montreal, Canada. His current
Engineering Prize in 2014, and the Postdoctoral Fellowship (Elevation research interests are microwave and millimeter-wave circuit design,
and acceleration) from Mitacs and FQRNT in 2017. hardware and software radio receivers, radars, and imaging systems.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on September 02,2021 at 08:00:38 UTC from IEEE Xplore. Restrictions apply.