A Terahertz Microscopy Technique for Sweat Duct Detection

Full text

A Terahertz Microscopy Technique for Sweat Duct Detection
Abstract— We propose a THz imaging technique that uses high
reflective index optics to improve spatial resolution and enable a
novel biometrics imaging tool. Specifically, with the use of THz
waves we can penetrate the drier outer skin layers and provide
anatomical information on the skin’s layered morphology and the
underlying structures (e.g. sweat ducts). Sweat ducts are
subcutaneous helical structures that exhibit absorption in the sub-
THz frequency range. The proposed THz microscopy
configuration can acquire high spatial resolution images of the
human skin and classify sweat ducts based on the backscattered
THz spectrum. In this paper, the theoretical background of the
microscopy technique and the experimental design are discussed.
Finally, THz images of human fingerprints are presented,
verifying the imaging capabilities of the proposed configuration.
Index Terms— Terahertz, microscopy, sweat ducts
I. THZ RADIATION IN SKIN IMAGING
Several applications that exploit the unique nature of THz
radiation have been developed over the last few decades.
Namely, the small wavelength of THz waves and their ability
to penetrate non-metallic structures, enable the design of
imaging systems that acquire high spatial resolution images and
extract the skin’s electromagnetic characteristics for
biomedical applications [1]-[5]. Specifically, a significant
research effort has been devoted in systems that acquire THz
images of cancerous tissue [1], since, its electromagnetic
properties can vary significantly from healthy tissue. These
systems are usually comprised of quasi-optical setups that
illuminate the tissue using a focused beam and measure the
reflected or transmitted signals. Similarly, THz radiation has
been used to image skin burns [1]. THz waves are sensitive to
the accumulated humidity underneath the wounded area,
therefore, they can be exploited for the accurate imaging of
burned tissue [1]. Apart from the attractive imaging properties
and although THz waves feature small wavelengths, image
spatial resolution is inadequate when needing to distinguish
smaller tissue traits.
In this work, we present imaging of the surface and
subsurface human skin traits using a high spatial resolution THz
microcopy technique. We device a quasi-optical configuration
consisting of a THz source, a paraboloid mirror and a high
resistivity silicon lens that enables the formation of a tightly
focused beam on the specimen surface. In this work, we are
mainly interested in the imaging of sweat duct patterns (helical
structures buried in the upper layers of skin), since, their
accurate detection could be exploited in diagnostic and
advanced security fingerprint biometrics.
Fig. 1. (a) Typical optical coherence tomography (OCT) image of the skin’s
cross-section showing the stratum corneum, sweat duct, epidermis, and dermis
[3]. (b) and (c) Numerical results of the reconstructed image at 400 GHz with
and without sweat ducts, respectively. Inset: Numerical model of the epidermis
featuring the top surface undulations (ridges and valleys) and several sweat
ducts.
Specifically, sweat duct patterns could serve as a tool in the
diagnosis of certain neuropathies (small fiber neuropathy and
diabetic neuropathy) by identifying their density and
functionality in the human skin [6]. Sweat ducts are helical
structures that outpour the sweat secreted in the sweat glands
which lie deeper into the skin. When certain neuropathic
diseases attack the small nerve fibers, sweat glands are
deactivated and inhibit sweat secretion. Even though, the effect
of this deactivation it still unclear, we expect that less sweat will
have direct effect on the conductivity of the ducts, thus,
affecting the electromagnetic response to THz radiation.
Additionally, the acquired sweat duct patterns can be used as an
extra security layer in fingerprint identification systems.
Namely, spoofing the surface fingerprints to bypass current
fingerprint scanners, is feasible nowadays with the
advancements in 3D printing technology [7]. However, sweat
ducts are subcutaneous structures that cannot be replicated from
surface residues of the human fingerprint, therefore, their
patterns can serve as an extra security layer in existing
identification systems. In the following sections, the theoretical
background of the proposed configuration and the experimental
setup are thoroughly discussed, along with a series of imaging
results.
II. SWEAT DUCTS AS A THZ HELICAL ANTENNAS
Several human skin traits (e.g. fingerprint undulations) and
sub-skin features (e.g. sweat ducts) are unique to every person
and remain unaltered during the course of their lifetime. As
depicted in an optical coherence tomography (OCT) image of
Fig. 1a, the human finger skin tissue has a rich morphology.
Namely, the skin can be modelled by three layers [3]-[5]: 1) the
Panagiotis C. Theofanopoulos, Student Member, IEEE and Georgios C. Trichopoulos, Member, IEEE
School of Electrical, Computer, and Energy Engineering, Arizona State University, Tempe, AZ, USA
Email:{ptheofan, gtrichop}@asu.edu

stratum corneum, 2) the epidermis, and 3) the dermis.
Additionally, the sweat ducts (diameter varies from 200-300
μm) are expected to behave as lossy helical antennas with a
broad resonance in the 200-500 GHz range [3]-[5].
TABLE I
ELECTROMAGNETIC PROPERTIES OF FINGER SKIN
Skin Trait
Permittivity (εr)
Conductivity (S/m)
Stratum Corneum
2.4
10-5
Epidermis
3.2
1
Dermis
3.9
30
Sweat Duct
4
100-10,000
Fig. 2 The proposed quasi-optical setup for the accurate human tissue imaging.
The electromagnetic properties of these skin traits are
summarized in Table I.
The effect of sweat ducts in the human skin’s reflectivity and
absorption of THz waves has been studied the past few years
[3]-[5]. Specifically, in [5], the absorption of THz waves in the
human skin, due to the presence of sweat ducts, was quantified
through full-wave simulations. The results presented in [5],
indicate that sweat ducts absorb THz waves significantly due to
high conductivity. Therefore, the effect of sweat duct
absorption should be detectable with the proper measurement
of reflected THz waves. However, previous studies use non-
focused imaging systems and the resonance effect of the sweat
ducts is masked by the lossy and anatomically complex
surrounding tissue. Here, we are proposing to scan the tissue
with a tightly focused THz beam and accurately identify the
sweat ducts from the rest of the skin traits.
Initially, to test the ability of focused THz radiation in human
skin imaging, we simulated the skin model shown on Fig. 1
(inset) using the tissue properties from Table I. The simulated
skin image is formed by raster scanning a tightly focused THz
Gaussian beam on the surface of the skin and by calculating the
reflected signals for each beam position. Figs. 1b and 1c show
the simulated THz reflection images for an undulated skin with
and without sweat ducts. Sweat ducts absorb the THz signal and
appear as darker spots along the friction ridges.
III. HIGH SPATIAL RESOLUTION IMAGING USING DIELECTRICS
WITH HIGH REFRACTIVE INDEX
In this section, the proposed THz microscopy configuration
is presented. The THz skin image is formed by raster scanning
a tightly focused Gaussian beam on the surface of the tissue
using x-y computer controlled translation stages and acquiring
the reflection coefficient for each beam position. The schematic
of the proposed setup is given in Fig. 2 and consists of a THz
transceiver (source and detector) operating at 325-500 GHz, a
paraboloid mirror and a hemispherical lens, placed on top of a
moving stage and a fixed stage that is supporting the sample
tissue. The beam of the THz source (diagonal horn antenna fed
by the VNA extender) is coupled to a paraboloid mirror and
then to a hemispherical silicon lens to form a converging
Gaussian beam that illuminates only a small spot in the surface
of the finger. The Gaussian beam-spot radius (wo) varies
approximately from 150-90 um for the 325-500 GHz band [8].
The use of the high reflective index silicon lens, enables high
spatial resolution imaging due to the formation of a spatially
small beam-spot compared to a free-space setup. The effect of
the high permittivity lens can be explained using the diffraction
limit equation of a microscope [9]
d 2nsin( )  
(1)
where, d is the minimum resolvable feature size, λ is the free-
space wavelength, n is the index of refraction of the medium
being imaged in, and θ is the half-angle subtended by the lens
aperture. Hence, by using a hemispherical silicon lens instead
of a free-space setup, we achieve ~3.42 better spatial resolution.
IV. THZ IMAGE MAGNITUDE AND PHASE CALIBRATION
The aforementioned imaging system constitutes a quasi-
optical setup that measures the reflection coefficient between
the silicon and an object that lies near the silicon surface.
Although the system can provide anatomical information with
high spatial resolution, spectral information can be also used to
improve skin trait classification. However, the raw reflected
spectra contains information related to the imaging setup
geometry due to multiple reflections and signal absorption
within the various components (waveguide, horn antenna,
mirror, and lens). Hence, the use of calibration is necessary to
accurately determine the spectrum of the reflection coefficient
that is related only to the specimen under imaging.
To account for the system’s measurement errors, we use the
3-term error model [10]. To accurately define the three error
terms of the quasi-optical setup, we use the following
calibration standards: 1) Silicon/Air interface with Γ=0.5505
[11], 2) a mirror placed on top of the silicon as a short with
Γ=−1, and 3) a shifted short using a 125 um thick quartz wafer.
As such, the reference plane for the THz images is shifted on
the dielectric len’s rear surface and the recorded image spectra
contains only the skin information.

V. THZ IMAGES OF THE HUMAN FINGERPRINT
The in-vivo THz image of a human fingerprint acquired
using our experiment setup is shown in Fig. 3a. Initially, a 3×2
mm area (40×30 pixels) is measured. In this image, the ridges
Fig. 3. (a) The acquired human fingerprint image at 400 GHz and (b) the
corresponding optical microscopy image.
Fig. 4. The calibrated spectra for three different positions on the THz fingerprint
image: a valley, a ridge and a sweat duct.
and the valleys of the fingerprint can be clearly identified by the
difference in the magnitude of the reflection coefficient.
Several sweat ducts can be identified as dark spots in the image
based on their THz absorption. In Fig. 3b the optical image of
human fingerprint is given, verifying the imaging accuracy of
the THz configuration. In the optical image, we can detect the
small openings of the sweat ducts (sweat pores) as white spots
on the ridges.
To improve the confidence in the classification of the sweat
ducts we also inspect the spectrum of every point/pixel on the
THz image. Figure 4 depicts the magnitude of the calibrated
reflected signals for different spots along ridges and valleys.
Specifically, three different positons are chosen: 1) a valley, 2)
a ridge, and 3) a sweat duct. As mentioned previously, the
valleys exhibit larger reflection than the ridges due to their
electromagnetic properties. However, in this image we can
observe the effect of the sweat ducts absorption spectra in the
reflection coefficient. Namely, the sweat duct introduces
greater losses in the reflected signal than the ridge, in the 380-
470 GHz region, enabling accurate identification based on the
absorption spectra.
VI. CONCLUSION
We have proposed a high spatial resolution microscopy
technique to acquire THz images of the human fingerprint for
the first time. The proposed configuration consists of a quasi-
optical scanner operating in the 325-500 GHz band. The main
advantage of the proposed setup is the use of a silicon lens to
focus a THz beam on the skin surface achieving ~3.42 times
better spatial resolution than the free-space case (εr=1). This
system can provide in-vivo human skin images enabling the
detection of the sweat ducts along with other surface skin traits
(e.g. ridges and valleys). Additionally, by calibrating the
reflection coefficients for each measured point of the
fingerprint, we can carry out a frequency analysis on the
calibrated data to extract features that are related with the
morphology of the sub-surface traits of the skin (e.g. sweat
ducts’ spectral resonance). Such THz images could serve as a
diagnostic tool for several neuropathic diseases and an extra
security layer for advanced fingerprint biometrics scanners.
Also, additional investigation is needed to properly characterize
the distinction between the active and inactive sweat ducts,
since the effect of the neuropathic diseases on the sweat duct
electromagnetic properties is still unclear. Finally, the proposed
setup can be modified to provide simultaneous dual-
polarization images of the human skin that can be exploited to
map the sweat duct patterns based on their exhibited dichroism
effects.
REFERENCES
[1] Sayuri Yamaguchi, Yasuko Fukushi, Oichi Kubota, Takeaki Itsuji,
Toshihiko Ouchi, and Seiji Yamamoto "Brain tumor imaging of rat fresh
tissue using terahertz spectroscopy," Nature, Scientific Reports 6, 30124,
2016.
[2] Tewari P, Kealey CP, Bennett DB, Bajwa N, Barnett KS, Singh RS, Culjat
MO, Stojadinovic A, Grundfest WS, Taylor ZD., “In vivo terahertz
imaging of rat skin burns,” Journal of Biomedical Optics, 17(4), 2012.
[3] Saroj R. Tripathi, Eisuke Miyata, Paul Ben Ishai, and Kodo Kawase,
"Morphology of human sweat ducts observed by optical coherence
tomography and their frequency of resonance in the terahertz frequency
region," Scientific Reports 5, Article number: 9071, 2015.
[4] Hayut, A. Puzenko, P. Ben Ishai, A. Polsman, A. J. Agranat and Y.
Feldman, "The Helical Structure of Sweat Ducts: Their Influence on the
Electromagnetic Reflection Spectrum of the Skin," in IEEE Transactions
on Terahertz Science and Technology, vol. 3, no. 2, pp. 207-215, March
2013.
[5] N. Betzalel, Y. Feldman and P. B. Ishai, "The Modeling of the
Absorbance of Sub-THz Radiation by Human Skin," in IEEE
Transactions on Terahertz Science and Technology, vol. 7, no. 5, pp. 521-
528, Sept. 2017.
[6] A. Hovaguimian and C. H. Gibbons. “Diagnosis and Treatment of Pain in
Small Fiber Neuropathy,” Current pain and headache reports 15.3 (2011).
[7] B. Tan and S. Schuckers, “Spoofing protection for fingerprint scanner by
fusing ridge signal and valley noise, In Pattern Recognition,” Volume 43,
Issue 8, 2010.

[8] P. F. Goldsmith, Quasioptical Systems: Gaussian Beam Quasioptical
Propogation and Applications, 1st ed., Piscataway, NJ, USA: Wiley-
IEEE Press, 1998.
[9] M. Born and W. Emil, Principles of Optics, Cambridge University Press,
1997.
[10] R. B. Marks, “Formulations of the basic vector network analyzer error
model including switch-terms,” in 50th ARFTG Conf. Fall Dig., Dec.
1997, vol. 32, pp. 115–126.
[11] S. J. Orfanidis, Electromagnetic Waves and Antennas, 2016.