Content uploaded by Anuj Baskota
Author content
All content in this area was uploaded by Anuj Baskota on Jan 19, 2024
Content may be subject to copyright.
Characterization of cells and tissues using a compact
GHz ultrasonic imager
Anuj Baskota, Justin Kuo, Serhan Ardanuç, and Amit Lal
Geegah Inc,
Ithaca, USA
anuj@geegah.com, justin@geegah.com, serhan@geegah.com, and amit@geegah.com
Abstract— Imaging technologies contribute to advances in
medical diagnostics and treatment of diseases by facilitating the
study of the structure and dynamic behaviors of cells and tissues.
This work reports on the use of a monolithic CMOS ultrasonic
imager as a platform for visualizing tissues and cells at GHz
frequency. The 128 x 128 array of AlN transducers allows for real-
time visualization of cells/tissues in dark environments, in addition
to the measurement of their mechanical properties.
Keywords—ultrasound, GHz ultrasonics, cell, tissue, acoustic
impedance, imaging
I. INTRODUCTION
Advances in imaging technology directly influence the
study of cells and tissues. The visualization of biological
features at a cellular level enables discovery of cellular
interactions and events within various types of tissues. High-
resolution scanning electron microscopy, near-infrared optical
imaging, and hyperspectral imaging allow highly specialized
imaging of cells up to a genetic level [1,2,3]. The dependency
of these imaging modalities on light/electron generating
sources results in bulky and expensive apparatus that limit the
mobility of the system and therefore limit the situations in
which they can be used. One alternate option that can
potentially solve this issue is to turn to use an imaging modality
that captures the mechanical properties of cells and issues. In
recent years, atomic force microscopy (AFM) has been
explored to study cell/tissue topography and their mechanical
properties [4]. Although AFM can yield high resolution images,
limitations include long acquisition time as well as intensive
sample preparation [5,6]. Common sample preparation required
for various imaging techniques include fixing samples on a
substrate, staining using fluorescence dyes and nano particles,
dehydrating, and thin film deposition of the samples which can
also be expensive, challenging, and extremely time consuming
[7,8,9]. Improper sample preparation can lead to damage of
biological samples and failure of obtaining high quality
microscopy images. In addition, to achieve real-time imaging
so that a doctor can receive instant feedback during a procedure,
this sample preparation step must be eliminated.
Another way to image mechanical properties of tissue
in a non-invasive way is to use ultrasound. Conventional
medical ultrasound in the 3-10 MHz frequency range is a
commonly used imaging technology for medical screening,
diagnosis, and treatments. The images obtained have a spatial
resolution of 0.15 to 5mm as wave frequency is inversely
proportional to the wavelength [10,11]. Although these are used
for visualization of large tissues and organs, research related to
using high frequency ultrasound to visualize and study cells is
still limited.
To solve these issues of optical microscopy requiring
sample preparation, of atomic force microscopy requiring large
and bulky systems, and of conventional medical ultrasound not
having high enough resolution, we propose instead to use a
GHz ultrasonic imager. In this paper, we present initial data on
using a GHz ultrasonic imager as a “platform” to study cells
and tissues. The GHz ultrasonic imager used in this work
consists of an array of 128 by 128 ultrasonic pixels that transmit
and receive ultrasonic waves. Each pixel comprises an
aluminum nitride (AlN) transducer and a transmit and receive
(T/R) circuit situated directly underneath the pixel transducer,
Figure 1. A) An array of 3 x 3 pixels with integrated CMOS T/R circuity. B)
Schematic showing ultrasonic imaging of onion epidermal cells on silicon
surface. C) The imager system with silicon surface as the sensing surface.
2023 IEEE International Ultrasonics Symposium (IUS) | 979-8-3503-4645-9/23/$31.00 ©2023 IEEE | DOI: 10.1109/IUS51837.2023.10307657
Authorized licensed use limited to: Cornell University Library. Downloaded on January 19,2024 at 20:17:22 UTC from IEEE Xplore. Restrictions apply.
as shown in Fig. 1A. The imager is fabricated using a
monolithic AlN on CMOS process.
The principle of operation for imaging is as follows.
An ultrasonic pulse with GHz carrier frequency (1.0 – 2.0 GHz)
is transmitted from an AlN transducer within a pixel. The
ultrasound propagates through the CMOS metal and dielectric
layers, and into the silicon substrate. When the ultrasound
reaches the opposite side of the silicon substrate, it is reflected
at that interface, as shown in Fig. 1B. The amount of ultrasound
that is reflected is determined by the reflection coefficient of
the material present at that interface with respect to silicon. This
interface therefore acts as the sensing surface for the imager.
The reflected ultrasonic wave then propagates through the
silicon substrate until it reaches the pixel transducers, where the
ultrasound is received and converted to an electrical signal. The
transit time for the wave to be transmitted and received by the
transducers is ~ 176 ns for each pixel. The samples placed on
the sensing surface can be acquired at a rate of up to 20 fps for
the entire 128 x 128 pixels. The spatial resolution achieved by
this imager is 50 , whereas the axial resolution is in the order
of several microns. The overall packaging of the GHz ultrasonic
imager is compact and handheld, with a silicon surface as the
sensing surface as shown in Fig. 1C. Unlike optical imaging
modalities using bulky light sources, the generation of
ultrasound using piezoelectric thin film MEMS transducers that
are monolithically integrated directly onto a CMOS chip results
in a small form factor. The biopsied samples of tissues can be
directly placed on the silicon surface right after extraction,
requiring minimum pre-processing. In addition, samples can be
visualized in the dark enabling study of cells/tissues sensitive
to light. The acoustic data obtained provides information about
the mechanical properties of tissues and cells, in addition to the
visualization aspect.
The feasibility of the GHz ultrasonic imaging
technology described in this work is shown by imaging several
samples of tissues and cells. As discussed in the next sections,
these samples and imaging experiments include the imaging of
chicken tissues, demonstration of contrast between fat and
muscle cells, the imaging of individual onion epidermal cells,
and the measurement of the acoustic impedance of different
meat samples (chicken, beef, tuna, salmon) along with several
onion epidermal cells.
II. METHODS
A. Preparation of tissue biopsies
Thin slices of tissues of chicken (muscle, fascia, and fat)
were extracted using a biopsy needle from commercially
purchased fresh meat packs. Muscle tissue samples were also
biopsied from tuna, salmon, and beef using the same method.
The thickness of these slices varied from 100 µm to 250 µm.
The moist samples were gently tapped using paper tower to
remove access water from their surface.
The biopsied samples for chicken tissues are shown in Fig.
2A. The imaging was performed for 10 replicates for each
chicken tissue, and 3 replicates for tuna, salmon, and beef.
B. Imaging of cells and tissues
The tissue samples were directly placed on the sensing
surface of the imager chip and slowly pressed using a glass slide
attached to a linear Z-stage (Thorlabs, MTS25-Z8) as shown in
Fig. 2B. This was done to ensure good contact between the tissue
and the silicon surface. Similarly, a thin layer of onion cell
membrane was extracted using a tweezer and placed on a glass
cover slip. This was immediately turned upside down to the
imaging surface (with onion membrane facing the silicon) to
avoid evaporation of water from the membrane. No additional
coupling media such as ultrasonic gel or water was used during
imaging.
The tissue samples and the onion cell membranes were
imaged at 1.853 GHz at room temperature of 20 ºC. The whole
frame (128 x 128 pixels) acquisition rate was 7 fps. The number
of frames acquired for tissue sample and onion cells was 100 and
250 respectively. An optical microscope camera (Hayear HY –
2307) was placed on top to visualize cells and tissues while the
ultrasonic images were being captured from the bottom. The
optical images were used to validate the spatial measurements
of the onion cells extracted from the ultrasonic images.
III. RESULTS AND DISCUSSION
The ultrasonic signals received by the 128x128 transducer
pixels are output from the system as demodulated in-phase (I)
Figure 2. A) Chicken tissues and biopsied samples. B) Schematic showing
the pressing of thin tissue slices/membrane layers using linear stage for
GHz ultrasonic imaging.
Authorized licensed use limited to: Cornell University Library. Downloaded on January 19,2024 at 20:17:22 UTC from IEEE Xplore. Restrictions apply.
and quadrature (Q) matrices. The first ultrasonic echo that is
received from the silicon-sample interface is captured as Iecho
and Qecho. A baseline voltage reading is measured after all the
ultrasonic echoes die off – this data is captured as Inoecho and
Qnoecho. This “no-echo” signal is used to remove DC offsets that
stem from the quadrature demodulation circuitry. Similarly,
another baseline measurement is taken when only air is in
contact with the sensor surface. This air baseline data is
subtracted from the measured echo data to visualize the
ultrasonic images with higher contrast.
The ultrasonic images (Q) for onion epidermal cells along
with the optical image is shown in Figure 3. The individual cells
are clearly visible due to the proper contact of cytoplasm on the
silicon surface. The spatial data of the cells is extracted from the
ultrasonic image that was interpolated using a bilinear filter.
Among the 18 cells analyzed, the length and width are measured
and a minimal variation in spatial data measurements between
the two imaging modalities – ultrasonic and optical – was found
as reported in our previous work on cell imaging [13].
The acoustic impedance (Z) is the resistance to the acoustic
waves by a material and can be defined as the product of density
() and speed of sound () of that material:
= ()
The echo and no-echo signals for both I and Q obtained from the
imager are used to compute the magnitude and phase of the
ultrasonic echoes. Furthermore, the reflection coefficient of the
sample with respect to the silicon layer (ΓSample/Si) is derived
from the magnitude of the air and sample. This coefficient refers
to the total energy of the acoustic wave that was reflected to the
transducers from the sample. The ΓSample/Si can be used to derive
the acoustic impedance of the sample using the following
relationship:
=
()
where ZSi is the acoustic impedance of silicon (~19.6 MRayls)
and is derived with the measurements from the
imager. The derivation from I, Q (echo and no-echo) matrices
to the Z of a sample is shown in our previous work [14].
The Z measured for chicken tissue slice with fat and
muscle layers is shown in Fig. 4A. Muscle cells being stiffer
absorb more ultrasound compared to the fat cells, therefore
have greater Z [15]. The distribution of measured Z of tissues
and cells is shown in Fig. 4B. The acoustic impedance recorded
for epidermal cells of onion is 1.53 0.087 MRayls (n = 18).
The obtained Z values for tissues fall under the same range as
reported in other literature [16,17,18]. The miniature size of this
Figure 3. The ultrasonic imager’s out of phase echo signal (units in mV)
and the optical image of the onion membrane showing individual cells.
Figure 4. A) Acoustic impedance 2D map of chicken tissue showing contrast between muscle and fatty regions. B) Acoustic impedance values measured
for chicken tissues (muscles, fat, and fascia: n = 10 samples: 100 frames per experiment), salmon, tuna, beef (n = 3 samples: 100 frames per experiment),
and onion cells (n = 18 samples: 250 frames per experiment).
Authorized licensed use limited to: Cornell University Library. Downloaded on January 19,2024 at 20:17:22 UTC from IEEE Xplore. Restrictions apply.
chip allows for integration surgical tools such as tweezers or
scalpels enabling tissue detection during surgical procedures.
This work demonstrates that a GHz ultrasonic imager
chip can be used to measure the acoustic impedance of tissues,
to resolve individual onion epidermal cells without the use of
stains, and to differentiate between fatty and muscle tissue in
chicken, thereby showcasing the potential of a CMOS
integrated high frequency ultrasonic imager as a compact and
powerful tool to visualize and study cells and tissues. The
imaging protocol involving direct contact of biopsied samples
on sensing surface eliminates the difficulties of tissue sample
preparation associated with other imaging modalities.
Without the need for sample prep such as staining,
imaging can be performed in real time and in situ. Therefore,
innovative integration of a GHz ultrasonic imager chip to
surgical instruments such as forceps can potentially enable
differentiation of different types of tissues in real time during
an operation. Such a system will benefit from further
miniaturization of the GHz ultrasonic imager system towards a
single module or a single chip solution. Integration of
peripheral circuitry such as ADCs, control logic, and PLLs onto
the imager chip would be required to realize this vision.
Along with the visualization of tissues and cells, the
measurement of acoustic impedance enables further study of
biomechanics of cells during growth and motility. Future work
on this front includes reducing the pixel size to achieve imaging
of mammalian and bacterial cells.
ACKNOWLEDGMENTS
This work was performed in part at the Cornell NanoScale
Facility, a member of the National Nanotechnology Coordinated
Infrastructure (NNCI), which is supported by the National
Science Foundation (Grant NNCI-2025233).
REFERENCES
[1] I. Mogollón and L. Ahtiainen, “Live Tissue Imaging Sheds Light on
Cell Level Events During Ectodermal Organ Development,” Frontiers in
Physiology, vol. 11, Jul. 2020, doi:
https://doi.org/10.3389/fphys.2020.00818
[2] L. Shemilt, A. Katrina, M. Yusuf, and I. Robinson, “Scanning electron
microscope studies of human metaphase chromosomes,” Philosophical
Transactions of the Royal Society A, vol. 372, no. 2010, pp. 20130144–
20130144, Mar. 2014, doi: https://doi.org/10.1098/rsta.2013.0144.
[3] N. Mehta, S. Shaik, R. Devireddy, and M. Ranjan Gartia, “Single-Cell
Analysis Using Hyperspectral Imaging Modalities,” Journal of
biomechanical engineering, vol. 140, no. 2, Jan. 2018, doi:
https://doi.org/10.1115/1.4038638.
[4] P. K. V. Babu, and M. Radmacher, “Mechanics of Brain Tissues Studied
by Atomic Force Microscopy: A Perspective,” Frontiers in Neuroscience,
vol. 13, Jun. 2019, doi: https://doi.org/10.3389/fnins.2019.00600.
[5] Y. F. Dufrêne, “Atomic Force Microscopy, a Powerful Tool in
Microbiology,” Journal of Bacteriology, vol. 184, no. 19, pp. 5205–5213,
Oct. 2002, doi: https://doi.org/10.1128/jb.184.19.5205-5213.2002.
[6] H.-J. Butt et al., “Scan speed limit in atomic force microscopy,” Journal
of Microscopy, vol. 169, no. 1, pp. 75–84, Jan. 1993, doi:
https://doi.org/10.1111/j.1365-2818.1993.tb03280.x.
[7] A. K. Pathan, J. M. Bond, and R. E. Gaskin, “Sample preparation for
scanning electron microscopy of plant surfaces—Horses for courses,”
Micron, vol. 39, no. 8, pp. 1049–1061, Dec. 2008, doi:
https://doi.org/10.1016/j.micron.2008.05.006.
[8] K. E. Kirat, I. Burton, V. Dupres, and Y. F. Dufrêne, “Sample preparation
procedures for biological atomic force microscopy,” Journal of
Microscopy, vol. 218, no. 3, pp. 199–207, Jun. 2005, doi:
https://doi.org/10.1111/j.1365-2818.2005.01480.x.
[9] W. M. Hess, “ Fixation and Staining of Fungus Hyphae and Host Plant
Root Tissues for Electron Microscopy”, Stain Technology, vol. 41 no. 1,
pp. 27-35, doi: 10.3109/10520296609116276
[10] A. Carovac, F. Smajlovic, and D. Junuzovic, “Application of Ultrasound
in Medicine,” Acta informatica medica : AIM : journal of the Society for
Medical Informatics of Bosnia & Herzegovina : asopis Drutva za
medicinsku informatiku BiH, vol. 19, no. 3, pp. 168–168, Jan. 2011, doi:
https://doi.org/10.5455/aim.2011.19.168-171.
[11] E. Fabiszewska, K. Pasicz, I. Grabska, W. Skrzyński, W. Ślusarczyk-
Kacprzyk, and W. Bulski, “Evaluation of Imaging Parameters of
Ultrasound Scanners: Baseline for Future Testing,” Polish Journal of
Radiology, vol. 82, pp. 773–782, Jan. 2017, doi:
https://doi.org/10.12659/pjr.904135
[12] A. M. Handorf, Z. Yun, M. A. Halanski, and W.-J. Li, “Tissue Stiffness
Dictates Development, Homeostasis, and Disease Progression,”
Organogenesis, vol. 11, no. 1, pp. 1–15, Jan. 2015, doi:
https://doi.org/10.1080/15476278.2015.1019687.
[13] A. Baskota, J. Kuo, S. Ardanuc, and A. Lal, “Compact GHz Ultrasonic
Micro-Imager for Cells and Tissues,” Microscopy and Microanalysis, vol.
29, no. Supplement_1, pp. 1116–1117, Jul. 2023, doi:
https://doi.org/10.1093/micmic/ozad067.572.
[14] A. Baskota, J. Kuo, S. Ardanuc, and A. Lal, “Nematode Species
Differentiation Using GHz Ultrasonic Micro Imager,” 2022 IEEE
International Ultrasonics Symposium (IUS), Oct. 2022, doi:
https://doi.org/10.1109/ius54386.2022.9957522.
[15] D. R. Wagner, “Ultrasound as a Tool to Assess Body Fat,” Journal of
Obesity, vol. 2013, pp. 1–9, Jan. 2013, doi:
https://doi.org/10.1155/2013/280713.
[16] G. D. Ludwig, “The Velocity of Sound through Tissues and the Acoustic
Impedance of Tissues,” Journal of the Acoustical Society of America, vol.
22, no. 6, pp. 862–866, Nov. 1950, doi:
https://doi.org/10.1121/1.1906706.
[17] T. Douglas Mast, “Empirical relationships between acoustic parameters
in human soft tissues,” Acoustics research letters online, vol. 1, no. 2, pp.
37–42, Oct. 2000, doi: https://doi.org/10.1121/1.1336896.
[18] T. Shishitani, S. Yoshizawa and S. -i. Umemura, "Change in acoustic
impedance and sound speed of HIFU-exposed chicken breast muscle,"
2010 IEEE International Ultrasonics Symposium, San Diego, CA, USA,
2010, pp. 1384-1387, https://doi.org/10.1109/ULTSYM.2010.5935709
Authorized licensed use limited to: Cornell University Library. Downloaded on January 19,2024 at 20:17:22 UTC from IEEE Xplore. Restrictions apply.