Treasures @ UT Dallas
Not a member yet
7697 research outputs found
Sort by
Demonstration of POC Biosensor Toward Clinical Translation for Patient Bed-side Monitoring
No full textThe research presented in this dissertation focuses on developing and characterizing a
multiplexed affinity based electrochemical biosensing device toward clinical translation. The
goal of this work is to establish a portable POC device for early disease detection across diverse
healthcare applications using low sample volume, rapid response time and usability amongst
minimally trained individual relying on ASSURED (Affordable, Sensitive, Specific, User
friendly, rapid, and Robust, Equipment free and Deliverable to end users) criteria. Primarily,
we designed a robust, non-faradaic electrochemical affinity biosensing platform for the rapid
assessment of parathyroid hormone (PTH) as a single biosensing system. Unique high density
semiconducting nanostructured arrays on a flexible sensing surface were used to create the
analytical nanobiosensor. The surface modification technique was specifically designed to
improve the interaction of the nanostructure–biological interface to capture the desired PTH
level in HS and plasma. This was followed by evaluating the analytical performance of the
developed biosensor with clinical rigor. The assay validation results were compared with
laboratory standard as reference with results that demonstrated comparable performance with higher accuracy. Next, the scope of the biosensor was expanded to solve a clinically
challenging problem of detecting host immune markers for life-threatening sepsis infection.
Herein, we demonstrate a first-of-a-kind multiplexed POC biosensing device that
simultaneously detects a panel of eight key immune response cytokine biomarkers in sample
volume equivalent to two drops of plasma and whole blood within 5 minutes without sample
dilution. Moreover, this work focuses on validating the developed biosensing device
with LUMINEX standard reference method for clinical translation using nearly 200 patient
samples. The DeTecT (Direct Electrochemical Technique Targeting) Sepsis biosensing device
is surface engineered with specific capture probes that utilizes EIS to measure the capacitive
impedance change reflecting binding interactions between the capture probe and target
biomarker enabling multiplexed detection. Specificity of the biosensor was validated using
cross-reactive studies, which displayed insignificant interference from non-specific
biomarkers. The biosensor also displays stable and repeatable performance. The novelty
presented in this research combines the effectiveness of choosing specific host immune
response biomarkers for detection of sepsis combined with unique surface modification
strategy coupled with EIS technique to enable efficient clinical decision-making process. This
unique sensor technology would allow medical practitioners to facilitate targeted interventions
for septic patients as a rapid prognostic approach, preventing complications arriving from
sepsis
Continual Learning With Applications in Different Scenarios
No full textDuring the past few years, Artificial Intelligence (AI) has experienced significant development
and dramatically affected our lives. Machine learning, as one of the fundamental sub-areas
in AI, attracts increasing attention from researchers from various domains. Thanks to the
rapid growth of advanced hardware and the exponential expansion of training data, deep
learning, which was prohibited due to its resource consumption, has become practical and
outperforms human beings in many areas. Nevertheless, it’s worth noting that static models
are sub-optimal solutions for a fast-changing digital world. Vast amounts of data and novel
concepts are being created every moment, yet it’s unrealistic to train a new ML model from
scratch to capture these items, in terms of both time and computational complexity. It’s not
hard to see that a continuously updating model is preferred over a one-time trained model.
As a consequence, continuous learning has been the focus of many researchers from different
domains.
Namely, continual learning concentrates on training a model with novel data continuously.
Examples of new data include data from shifted data distribution, novel classes, and new
domains. One main challenge of continual learning is catastrophic forgetting, which refers
to forgetting acquired knowledge as a result of learning new concepts. This problem is more
severe in deep learning since most deep learning models are parameter-sensitive. On the
other hand, limited storage is another major blocker. It’s impractical to save all the historical
and newly emerging data simultaneously for model training. Moreover, this limitation also
contributes to catastrophic forgetting.
Continual learning is a broad research area. We propose several frameworks to address chal-
lenges in continual learning from different angles. We introduce Continual and Interactive
Feature Distillation for Multi-Label Stream Learning (CIRDM) using knowledge distillation
and label correlation for multi-label classification in continual learning. For the text in-
cremental learning, we propose a dual contrastive learning (DCL) based framework, which
produces embeddings for general knowledge and attention vectors for task-specific informa-
tion. It demonstrates the exceptional transferability of knowledge across different tasks. It
is important to uncover data from novel classes and classify data from known classes, Co-
Representation Learning for Open-Set Classification (RLCN) is introduced to address this
problem. Following the motivation of disentangling task-agnostic and task-specific knowl-
edge, we integrate prompt learning into continual learning. Improving the quality of stored
data is also a technique in continual learning. We combine these two ideas to solve the text
incremental learning problem.
Based on our research, we demonstrated the applicability of continuous model updating, and
its capability to avoid catastrophic forgetting in various scenarios.
Based on our research, we demonstrated the applicability of continuous model updating, and its capability to avoid catastrophic forgetting in various scenarios
Nano-biothiol Interactions of Engineered Nanoparticles
No full textNanomedicines have been extensively studied in the past decades at the fundamental level
because they could potentially make a paradigm shift in human healthcare. Nano-bio interactions
play a central role in the precise control of the benefit and hazards of nanomedicines, but current
studies mainly focus on how nanoparticles are taken up by cells and interact with different
receptors. There is still not enough investigation of how the physiological environment
transforms engineered nanoparticles through a variety of biochemical reactions. This dissertation
aims to fundamentally understand the nanoparticle-biochemical interactions and the in vivo
transport of engineered nanoparticles modulated by these interactions.
In Chapter 1 of this dissertation, an overall review is given on the current understanding of nanobio interactions at the molecular and chemical levels, particularly. In Chapter 2, we
systematically investigated how the nanoparticle size, the thiols species, and the protein binding
affect the interactions between the nanoparticles and thiols at the in vitro level. In Chapter 3, we
focused on unraveling the relation between the nanoparticle-biothiol interactions in vitro and the
nanoparticle-biothiol interactions in vivo. In Chapter 4, we explored the nanoparticle-biothiol
interactions in the diseased mice model and illustrated the application of nanoparticle-biothiol interactions in disease diagnosis. Finally, in Chapter 5, we present the summary and outlook.
These new understanding on nano-biochemical interactions at both in vitro and in vivo levels will
help further advance physiology at the nanoscale as well as open new pathways to early disease
diagnosis and treatment
Investigating the Role of Patterned Tissue Stiffness, Cell Proliferation and YAP Localization in Embryonic Kidney Development
No full textDuring renal development, the metanephric kidney arises when the ureteric bud forms along the
Wolffian duct and undergoes a series of branching events to build the ureteric tree. The tips of this
tree interact with renal vesicles in the metanephric mesenchyme to induce the formation of
developing nephrons and later fuse with them to form the filtration system within the kidney.
Kidney development, is an understandably a complex process, regulated in parts by GDNF/Ret
and Wnt signaling. A well-formed network of collecting ducts is essential for normal kidney
functioning as defects in branching morphogenesis are thought to be associated with chronic
kidney diseases. Proper renal branching morphogenesis depends crucially on cell proliferation,
and it is observed that proliferation is elevated specifically at the tips of branching ureteric tree.
Whether or not proliferation is developmentally patterned within the developing kidney and what
regulates this pattern of proliferation, is poorly understood. Although tissue mechanics has been
shown to influence the morphogenesis of other branched organs, such as the lung and mammary
gland, it is unclear how biophysical factors within the embryonic kidney, such as tissue stiffness
and cell proliferation, affect renal development and, how changes in the mechanical environment
in the embryonic kidney might interact with signaling cascades, such as the Hippo pathway, or
those downstream of GDNF, Wnt, and TGF-β, which are known regulate renal branching
morphogenesis. In this work, we quantified regional differences in tissue stiffness within the
embryonic kidney and investigated how these variations influence the patterns of proliferation that
sculpt the ureteric tree. We also investigated the effect on branching morphogenesis when patterns
of proliferation and Yap localization were disrupted. Taken together our data will help elucidate
the role and regulation of patterned biophysical factors within the embryonic kidney and its effect
on branching morphogenesis. These findings will further our understanding of branching-related
kidney diseases and can provide better insight towards tissue engineering efforts of rebuilding a
kidney
Gravitational Lensing of Gravitational Waves: Effects of Different Lens Models
No full textStrong gravitational lensing of gravitational waves (GWs) occurs when the GWs from a
compact binary system travel near a massive object. The lensed waveform is given by
the product of the lensing amplification factor F and the unlensed waveform. For many
axisymmetric lens models such as the point mass and singular isothermal sphere that we
consider, F can be calculated in terms of two lens parameters, the lens mass ML and source
position y. In the geometrical-optics approximation (GO), lensing in these models produces
at most two discrete images which can be parameterized by two image parameters, the flux
ratio I and time delay ∆td between images. In the macrolensing regime for which ∆td is
large compared to the time T they spend within the sensitivity band of GW detectors, it is
natural to parameterize lensing searches in terms of these image parameters. The functional
dependence of the lensed signal on these image parameters is far simpler, facilitating data
analysis for events with modest signal-to-noise ratios, and constraints on I and ∆td can be
inverted to constrain ML and y for any lens model. Previously unexplored, we use image
parameters to determine the detectability of gravitational lensing of GW in the microlensing
regime (∆td ≪ T ) and find that for GW events with signal-to-noise ratios ρ and total
mass M , lensing should in principle be identifiable for flux ratios I ≳ 2ρ−2 and time delays
∆td ≳ M −1.
We also study GW lensing using non-axisymmetric lens models, such as the singular isothermal ellipsoid (SIE) lens. The use of SIE lens model is motivated by observational constraints
on the shapes of galaxies and their dark matter halos, as well as the existence of EM strong
lensing configurations with more than 2 images. An SIE can produce four images when the
source and lens are well aligned in projection, the cross-section for which depends on the
lens ellipticity. In case of the SIE lens, the GO becomes invalid when the source position is
near the non-analytic regions, such as the cusps and folds, of the source plane. Therefore,
for the first time, the quasi-geometrical optics approximation (QGO) is employed in order
to quantify the breakdown of the GO. Using QGO we calculate the lower bound on the lens
mass, below which the GO is invalid. We also perform match-filtering analysis between GW
sources (lensed by an SIE, producing four images) and (i) unlensed templates and (ii) two
types of two-image lensed templates, where the images have the same or different Morse
index. Analogous to axisymmetric lens models, we use image parameters to derive analytical expressions to explain the mismatch behaviors for different choices of templates. These
investigations will have an impact on searches for GW lensing events in data from current
and future detectors.
Orbital precession is a dramatic effect induced by spin-orbit and spin-spin couplings in a compact binary source; this precession then imprints characteristic modulations in the waveform.
Search efforts are underway to detect GWs with precession during the inspiral phase of the
binary system. In this study, we quantify the precessional effects by introducing five new
phenomenological parameters. These parameters are expected to provide a comprehensive
understanding of the amplitude and frequency modulation, especially from the geometrical
perspective. We also analyse the mismatch between precessing source and non-precessing
templates to determine the minimum signal-to-noise ratio needed to detect precession for
upcoming observing runs of LIGO and third generation detectors
Rate and Performance Enhancement of LDPC Coded Schemes
No full textIn the first part of the dissertation, a novel collection of punctured codes decoding (CPCD)
technique that considers a code as a collection of its punctured codes is proposed. Two forms
of CPCD, serial CPCD that decodes each punctured code serially and parallel CPCD that
decodes each punctured code in parallel, are discussed. In contrast to other modifications
of Low-density parity-check (LDPC) decoding documented in the literature, the proposed
CPCD technique views a LDPC code as a collection of punctured LDPC codes, where all
punctured codes are derived from the original LDPC code by removing different portions
of its parity bits. CPCD technique decodes each punctured code separately and exchanges
extrinsic information obtained from that decoding among all other punctured codes for their
decoding. Hence, as the iterations increase, the information obtained in the decoding of
punctured codes improve making CPCD perform better than standard decoding. LDPC
codes have received significant interest in a variety of communication systems due to their
superior performance and reasonable decoding complexity. Numerical results demonstrate
that CPCD can significantly improve the performance, or significantly increase the code rate
of LDPC codes. It is demonstrated that both serial and parallel CPCD have about the same
decoding complexity compared with standard sum product algorithm (SPA) decoding. It
is also demonstrated that while serial CPCD has about the same decoding delay compared
with standard SPA decoding, parallel CPCD can decrease the decoding delay, however, at
the expense of processing power. Furthermore, it is demonstrated that similar improvements
in performance and decoding delay can be achieved by applying CPCD to longer codes with
higher-order modulation too. Specifically, it is shown that parallel CPCD with two parallel
concatenated codes D = 2 achieves 0.3 − 1 dB gain over standard SPA decoding of the
LDPC code of length 1944 employed in the WiFi standard with QPSK, 16-QAM or 64-
QAM modulation while simultaneously reducing decoding delay by about 50%. It is also
shown that the CPCD technique can similarly improve the performance of the LDPC code
employed in the 5G NR standard at its highest code rate by about 0.6 dB while reducing
the decoding delay by about 50%.
In the second part of the dissertation, a novel implicit transmission with bit flipping (ITBF)
technique is introduced to transmit a coded stream implicitly while transmitting a coded
stream explicitly over a channel. ITBF flips a set of chosen parity bits of the explicitly
transmitted stream according to an implicit stream. Numerical results show that the ITBF
can transmit an implicit stream at the rate up to 13.19% of the explicit stream without
significantly sacrificing performance, or increasing the decoding complexity or the decoding
delay. The ITBF is combined with CPCD to form ITCD schemes that can further increase
the rate of transmission on the implicit stream. It is demonstrated with the LDPC code in
the WiFi standard that ITCD can transmit an implicit stream at up to 25% of the rate of
the explicit
Sensorimotor Network Contributions to Rhythm, Syntax, and Domain-general Cognitive Processing
No full textRhythm and syntax have many behavioral, theoretical, and neural similarities that suggest that some of the underpinning neural resources may be shared. However, no work has sufficiently examined neural overlap within the same group of participants, nor has anyone examined contributions of beta oscillations—which are theorized to underpin predictive coding—to rhythm and syntax behavior using transcranial alternating current stimulation. This dissertation outlines three experiments that used functional magnetic resonance imaging and transcranial alternating current stimulation to identify potentially overlapping neural circuits recruited by both rhythm and syntax that are distinct from domain-general multiple demand processes. In the first experiment, the pre-supplementary motor area was identified as a candidate region that overlaps across rhythm and syntax but is distinct from multiple demand cognition. The second and third experiment examined how beta band pre-supplementary motor area activity contributes to syntax and rhythm behavior respectively. Participants who received targeted beta band transcranial alternating current stimulation to the pre-supplementary motor area responded more accurately to syntactically complex sentences compared to a group who received sham stimulation.
Furthermore, participants who received the same stimulation responded more quickly during a rhythm discrimination task compared to peers who received sham stimulation. These three experiments suggest that beta oscillations in the pre-supplementary motor area are possibly involved in both rhythm and syntax perception. I propose that these phenomena indicate that predictive coding of temporal events is the shared mechanism between rhythm and syntax, and I highlight additional experiments that can be performed to critically examine this theory
Quantum Transport in Josephson Junctions and Carbon Based Nano-materials
No full textIn topologically non-trivial band structure materials, the current distribution is of great
interest and several methods have been used to map the conducting channels. In this work,
we performed low-temperature quantum transport experiments in a Josephson Junction of
WTe2 to explore the edge and bulk contributions to the total conductivity in the sample.
Novel interference patterns were observed due to interplay of the bulk and edge conductance.
Graphyne, a new allotrope of Carbon has been theoretically predicted several years ago
but has been experimentally realized only recently. Theoretical modeling has shown this
material to manifest high mobility comparable to that of graphene but also have a finite
band gap making it an ideal candidate as transistor channel. Material quality is still under
being improved to access the intrinsic properties of this material. Several attempts have
been made to fabricate devices with this material but the yield has been low resulting in
limited preliminary data.
In another study, we attempted to fabricate superconducting wires using a new novel method.
Superconducting wires are of great importance in a number of applications ranging from
superconducting magnets to radiation shielding in spacecrafts. NbTi thin films are sputtered
on PVA (Poly Vinyl Acetate) which acts as a flexible substrate. NbTi films showed expected
superconducting properties. The sample is then rolled to form a wire and thereafter PVA is
eliminated by dissolving and NbTi shells coalesced to form a wire. This NbTi wire however
likely suffered damage during processing and lacked superconductivity. However, this sample
showed Two Level Fluctuations resembling Random Telegraph Noise signal which is studied
Bioinspired Wet Pressure-sensitive-adhesives and Shape Memory Polymers for Biomedical Applications
No full textNeural interfaces-integrated devices provide a promising technology enabling the diagnosis and treatment of neurological diseases and disorders by detecting and stimulating nervous action using bioelectronics. Nerve cuff electrodes are a class of neural integrated devices that consist of a flexible polymeric substrate that wraps a target nerve in cuff-shaped and single- or multi-channel electrodes that detect and/or stimulate nerves via excitation or inhibition of the target nervous action for neuromodulation. In this study, a new cuff-closing method is described, which uses shape memory polymer (SMP) cuffs and wet pressure-sensitive-adhesives (PSAs), for less invasive implantation. SMP cuffs wrap around a target nerve through shape recovery induced by exposure to physiological conditions, then fixed by wet PSA instead of suturing, to facilitate snugly fitting and easier and quicker setup for implantation. To demonstrate the feasibility of the proposed cuff devices, research was performed on three key scientific/technological issues, namely: 1) bioinspired wet PSA suitable for physiological conditions, 2) SMPs with enhanced softening and shape recovery properties, and 3) underwater adhesion of the designed SMPs and wet PSAs. First, the relationship between cohesion and adhesion of bioinspired catecholic PSAs, containing various amines, was studied to develop wet PSAs. The cation- interaction between catechol and amine groups contributed to the high cohesion of the PSAs and their high cohesion significantly increased adhesion under physiological conditions. Second, new thiol-ene/acrylamide SMPs were designed to improve shape recovery properties for facile implantation, using dopamine acrylamide (DAc) as a hydrophilic monomer. Finally, adhesion tests, for the designed wet PSAs to SMPs, demonstrated that the adhesion of PSAs to DAc-modified SMPs increased with an increase in the DAc molar ratio of SMPs to reach a sufficiently high ratio for permanent joints to be formed under physiological conditions. The closure method using wet PSA and SMP substrates, discussed in this study, is proposed as an advantageous method for designing minimally invasive nerve cuffs
Synthetic Gene Circuits as Benchmarks for Understanding Biological Networks
No full textThe expression of genes is controlled by regulatory networks, which perform fundamental information processing and control mechanisms in a cell. Unraveling and modelling these networks will be indispensable to gain a systems-level understanding of biological organisms and genetically related diseases. With their ability to emulate and interface with naturally occurring networks, synthetic gene networks are powerful tools in this process. Recent advancements in genetic engineering technologies have expanded the possibilities in design and implementation of synthetic networks, offering unprecedented opportunities to examine and perturb their activity in cellular milieu. In this thesis, we present development and characterization of synthetic gene circuits constructed specifically for the purpose of mimicking and monitoring the regulatory strategies in human cells. First, we introduce a reverse engineering pipeline using synthetic gene circuit as a benchmark for biological reverse engineering. We discuss the advantages of this method and how one can engineer the circuits amenable to reverse engineering. Using several synthetic gene circuits, we show that network reconstruction results not only reproduce the benchmark network topologies, but also identifies a novel feature that can be critical towards solving a commonly misidentified topology. Furthermore, we consider the
application of using synthetic circuits to characterize a rare and complex gene regulatory motif on its output expression. Specifically, we demonstrate that an intragenic miRNA-mediated output regulation operates as a filter with respect to promoter strength and reduces expression noise. Lastly, we present novel microRNA sensing systems based on CRISPR/Cas systems that offer new opportunities for engineering systems