31825 research outputs found
Sort by
Hybrid deep learning models for 2D computer vision
No full textKambhamettu, ChandraComputer vision has significantly evolved with the introduction of Vision Transformers (ViTs), which employ self-attention mechanisms to capture global context by considering relationships across the entire image. Previously, Convolutional Neural Networks (CNNs) were widely used, utilizing convolutional kernels to effectively capture hierarchical local spatial features such as textures, edges, and shapes. This thesis explores the integration of CNNs and ViTs into hybrid models to balance the local feature extraction capabilities of CNNs with the global context modeling abilities of ViTs, aiming to enhance performance in dense prediction tasks such as Salient Object Detection (SOD) and Edge/Contour Detection. ☐ This dissertation presents several innovative hybrid deep learning models that combine the strengths of CNNs and ViTs while overcoming their limitations. First, I present ConvSegFormer, a convolution-aided Vision Transformer that efficiently learns discriminative features from minimal data, competing well with state-of-the-art CNN architectures. Second, I present SODAWideNet, a hybrid deep learning model that incorporates self-attention into a convolutional architecture for SOD. It achieves competitive results without large-scale pre-training by using parallel attention and convolutional branches. Third, I present SODAWideNet++, a modified SODAWideNet model that merges the attention and convolution branches into a single branch, utilizing dilated convolutions for feature extraction. With a COCO-based pre-training pipeline, it delivers excellent performance with fewer trainable parameters. Fourth, I present SODDCNet, a CNN architecture that uses attention-generated convolutional weights to capture localized, instance-specific semantic features, achieving state-of-the-art results in SOD and video SOD. Fifth, I present USODFuseNet, an RGB-D SOD model that excels in underwater and RGB-D tasks, featuring a dual-branch encoder architecture with an efficient feature merging operation.University of Delaware, Department of Computer and Information SciencesPh.D
Non-invasive evaluation of in vivo intervertebral disc mechanical function
No full textElliott, Dawn M.The intervertebral discs are musculoskeletal soft tissues, situated between the vertebrae in the spine, that enable distribution of large, multiaxial loads as a part of daily activity and motion. Disc degeneration is the structural and compositional deterioration of the disc, with associated changes in mechanical properties and function. Disc degeneration is often implicated as a primary cause of low back pain (LBP). With normal aging degeneration there is disc breakdown that progresses steadily with increasing age. In accelerated degeneration, disc breakdown exceeds normal aging degeneration due to multifactorial, largely unknown causes. It has been notoriously difficult to distinguish between normal aging degeneration and accelerated degeneration. The relationship and associations between low back pain likely differ between aging degeneration and accelerated degeneration, such that assessing these relationships with no regard for subject age complicates study outcomes. ☐ Magnetic resonance imaging (MRI) is a non-invasive method commonly used to evaluate disc health, but structural and biomarker assessments from MRI have been largely unsuccessful in distinguishing between aging degeneration and accelerated degeneration. In the present work, baseline MRI parameters were evaluated for adults across the lifespan (n = 84, ages 18-83 years old) to quantify the impact of age on MRI outcomes. Given the known age-dependence of degeneration, we sought to develop a novel, age-dependent disc degeneration model. We developed a Disc Effective Age model, which requires subject traits and baseline MRI measures and outputs a Disc Effective Age. Following stepwise regression and assessment of several models, we found that a 4 parameter model optimally estimates the disc effective age. The model calculates disc effective age by disc level with inputs: subject height, nucleus T2 relaxation time, mid-sagittal disc area, and disc 3D volume. The model output can be directly compared to the subject’s true age to determine whether the degree of disc degeneration is expected due to aging or accelerated compared to asymptomatic subjects of the same age. The effective age model is objective, quantitative, continuous, and can be directly compared to a subject’s true age, enabling identification of discs with accelerated degeneration. ☐ Furthermore, the impact of aging degeneration on discs’ mechanical function in vivo is unknown. Another objective of this work was to evaluate disc mechanical function in vivo with repeated MRI in an asymptomatic adult population to quantify changes in disc mechanical function due to aging. MRI was acquired in flexion, extension, and diurnal loading conditions. Flexion and extension were applied as bending often inhibits or exacerbates LBP, such that these loading states have potential clinical relevance. Diurnal loading was evaluated as it is a natural, daily recurring loading state related to both disc loading and fluid flow. ☐ For many MRI parameters, the magnitude and distribution of outcomes were age-dependent. Generally, disc height and nucleus T2 relaxation times decreased while disc width and annulus T2 relaxation times increased with increasing subject age. Notably, in flexion, the changes in wedge angle and strain for younger subjects were greater in the lower lumbar discs whereas older subjects exhibited a greater flexion response in the upper lumbar discs. These findings highlight changes in disc mechanical function with age and can be further used as baseline expectations for evaluating mechanical response of subjects with LBP to help isolate potential sights of disc dysfunction. ☐ Another method for evaluating disc mechanical function is with ex vivo mechanical testing of cadaveric discs; unfortunately, it is unknown how directly these outcomes translate to the in vivo context. There are significant differences between the in vivo and the ex vivo states, including active loading, fluid flow, and surrounding tissues. It is necessary to quantify differences between the in vivo and ex vivo experimental states in order to accurately and effectively interpret ex vivo mechanical outcomes in the in vivo context. Given the necessity of paired samples between in vivo and ex vivo states, this quantification is only possible with the use of an animal model. We conducted sequential MRI on minipig spines throughout the dissection process to enable evaluation of the differences that arise between subsequent experimental states. The disc geometry, MRI T2 relaxation time, and opening pressure were all altered by progressive dissection, specimen preparation, and imposed loading conditions. The main finding of this study was that in the minipig model, an imposed axial stress of 0.20-0.33 MPa successfully recovered live, in vivo disc geometry and opening pressure. ☐ Besides mechanical testing, finite element modelling (FEM) can also be used to evaluate the ex vivo disc. FEM can evaluate disc response, isolate the impact of particular features, and provide internal stress and strain distributions, which are generally unrealistic to acquire with typical imaging and mechanical testing. The usefulness of a FEM is similarly limited by how well it can mimic the biological disc. To this end, the final objective of this work was to develop an intervertebral disc finite element model with geometry based on disc MRI and tissue properties from mechanical testing and evaluated its ability to mimic multiaxial disc mechanical function. We successfully incorporated residual strain due to swelling and multigeneration fibers in a FEM of intervertebral disc and validated human disc models against uniaxial quasi-static and multiaxial dynamic tests. The inclusion of swelling and fiber-induced residual strain in the multigeneration model was necessary for achieving a physiological residual strain state and for replicating disc mechanical behavior across uniaxial quasi-static and multiaxial dynamic test cases. ☐ The outcomes of this dissertation include a Disc Effective Age model which quantified aging degeneration, in vivo disc mechanical response across the adult lifespan with flexion, extension, and diurnal loading, quantified differences between the in vivo and ex vivo states in a porcine model, and validated an intervertebral disc finite element model with residual strain.University of Delaware, Department of Biomedical EngineeringPh.D
Social support and communicative disenfranchisement: extending the theory of motivated information management within a contested illness context
No full textCrowley, John P.The way that individuals with contested illnesses seek out information about their illness plays a major role in determining their quality of life. This study used the Theory of Motivated Information Management to examine how individuals with contested illnesses deal with uncertainty and the factors that influence the usage of information-seeking strategies and how that influences their quality of life. The results reveal that there is a negative relationship between avoiding information and quality of life for individuals with contested illnesses. The results also demonstrate that for individuals with low perceived social support, there is a negative relationship between direct information-seeking and quality of life whereas for individuals with high perceived social support, this relationship is positive. The results also show that for individuals with low perceived communicative disenfranchisement, there is a positive relationship between direct information-seeking and quality of life, whereas for individuals with high perceived communicative disenfranchisement, this relationship is negative.University of Delaware, Department of CommunicationPh.D
EFFECTS OF SMALL MOLECULE INHIBITORS ON MOTILITY AND PROLIFERATION OF UNDIFFERENTIATED, DIFFERENTIATED, AND IRRADIATED GLIOBLASTOMA STEM CELLS
No full textenterGlioblastoma (GBM) is an aggressive and incurable form of brain cancer.
Glioblastoma stem cells (GSCs) make GBM treatment difficult as they invade
surrounding brain tissue and are resistant to radiation and chemotherapy. Decreased
motility and proliferation of GBM cells from established glioblastoma cell lines was
previously observed when using small-molecule inhibitors fibroblast growth factor
receptor inhibitor PD173074, focal adhesion kinase inhibitors PF431396 and Y15, and
αvβ3/αvβ5 integrin inhibitor cilengitide. In this research it was found these inhibitors
were similarly effective at decreasing the motility of patient-derived GSCs. These
results were expanded to test the hypothesis that motility and proliferation of
“differentiated” and irradiated GSCs would be similarly inhibited. GSCs were cultured
in high-serum media to induce a differentiated state. Some of these “differentiated”
GSCs were later cultured in stem cell media. A “super scratch” assay measured the
velocity of migrating cells exposed to the inhibitors using time-lapse microscopy. A
proliferation assay was then performed to determine the percentage of cells in S phase
of the cell cycle. All cell lines exhibited similar decreases in motility following
inhibitor treatment. The “differentiated” GSCs exhibited slower overall motility, but
“differentiated” GSCs put back into stem-cell media exhibited similar motility as
regular GSCs pre-differentiation. The proliferation assay showed that “differentiated”
GSCs exhibited increased proliferation rates overall. The irradiated GSCs exhibited
similar responses to the inhibitors compared to non-irradiated GSCs in the motility
and proliferation assays. These results suggest the small molecule inhibitors tested
may be effective against both GSCs and differentiated GBM cells and provide
evidence that GSCs can enter in and out of a differentiated state depending on the
extracellular environment. Radiation also did not change the response to drug
treatments, which indicates radiation may not impact the efficacy of
chemotherapeutics during glioblastoma treatment.ente
THE C-TERMINAL DISORDERED REGION OF THE RNA HELICASE DEAD IS REQUIRED FOR COLD SHOCK GROWTH IN E. COLI
No full textenterDeaD, a protein belonging to the dead box family of RNA helicases in
Escherichia coli (E. coli), has many cellular functions. The known functions of DeaD
include its function as an RNA helicase, a role in ribosome biogenesis and enabling
growth at cold shock temperatures. DeaD expression is upregulated in cold
temperatures, which has led to it being given an alternative name: cold shock DeaD
box protein A (CsdA). We want to explore the function of DeaD during E. coli growth
under cold shock conditions. Most of the protein’s structure has been attributed to the
RNA helicase function. In addition to the RNA helicase domains, DeaD has two
intrinsically disordered regions (IDRs) and a DbpA domain. The function of these
other regions is less well researched. Using epifluorescence microscopy and growth
assays, I have been able to outline the significance of the C-terminal IDR to DeaD's
primary functions. In addition to assessing the function of DeaD as a cold shock
protein, we also want to characterize its localization within the cell.
Dead and other Deadbox helicases have been studied for their role in forming
RNA condensates. This phase-phase separation is important to cellular organization
and has been observed in both prokaryotic and eukaryotic cells. Since the C-terminal
IDR and DbpA domain have been implicated as RNA-binding domains, I wanted to
see if DeaD could form condensates without the C-terminal IDR. By fusing the DeaD
protein with a fluorescent protein, we can use microscopy to track protein localization
inside the cell. A deaD deletion strain of E. coli was transformed with tetracycline
inducible plasmids encoding either the full-length dead or the truncated form tagged
with a fluorescent protein. Full-length DeaD was able to form puncta at both preferred
temperature and cold shock, while the truncation was not entirely successful. Based on
our results, we have concluded that functioning DeaD is necessary for cells to survive
cold shock conditions and for this localization to occur.ente
Short-range order and longer-range disorder revealed in germanium–tin alloy thin films by extended x-ray absorption fine structure analysis Open Access
No full textThis article was originally published in Journal of Applied Physics. The version of record is available at: https://doi.org/10.1063/5.0245736.
© 2025 Author(s). All article content, except where otherwise noted, is licensed under a Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).Short-range order (SRO) in semiconductor alloys, a relatively under-studied structural phenomenon in which local atomic arrangements differ from those of a random solid solution, is investigated in molecular beam epitaxy (MBE)-grown GeSn thin films. A novel preparation technique is used to pattern these films into microscale ribbons that are released from the substrate for extended x-ray absorption fine structure (EXAFS) analysis. The results indicate a strong SRO in which the first shell around Sn atoms is greatly denuded of Sn atoms relative to the nominal atomic composition of the alloy. This effect is more pronounced than that observed recently in GeSn nanowires grown by chemical vapor deposition. Additionally, the presence of a longer-range disorder detected by EXAFS analysis in the shells of atoms more distant from the absorbers is indicative of the defects and inhomogeneous strain present in the MBE-grown films. The evident existence of the SRO in GeSn alloys deposited by different growth methods and in different strain states suggests that SRO is a general phenomenon in the thin films of this metastable solid solution.This research was supported by the U.S. National Science Foundation under Grant No. DMR-2003266. This funding primarily supported growth of the nanowires sample. This research was supported by the U.S. Department of Energy under Award No. DE-SC0023412 and Subaward No. UA2023-351. This funding primarily supported EXAFS characterization. The use of the Stanford Synchrotron Radiation Lightsource, SLAC National Accelerator Laboratory, was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences under Contract No. DE-AC02-76SF00515. Part of this work was performed at the Stanford Nano Shared Facilities (SNSF), supported by the National Science Foundation under Award No. ECCS-2026822. This research was supported by the U.S. National Science Foundation, NSF Fuse Program, Grant No. 2328840. This funding primarily supported MBE growth and sample preparation. Additional support was provided by the National Institute of Standards and Technology
MORE THAN POLITENESS: REIMAGINING CIVIL DISCOURSE IN THE FIRST-YEAR EXPERIENCE
No full textenterIn an era with increased political polarization and decreased democratic
participation, the need for civil discourse education, specifically in higher education, is
greater than ever. This thesis explores the impact of a short-term civil discourse
intervention on first-year honors students at the University of Delaware and
investigates scalable policy solutions to integrate civil discourse education into the
entire university population.
Through a quasi-experimental design, this study evaluates the effectiveness of
a Braver Angels parliamentary-style debate within a mandatory First Year Seminar
(FYS) course. A pre-post survey assessed changes in students’ attitudes and behaviors
related to six key variables: students’ views on the value of civil discourse, recognition
of knowledge limitations, empathy for different perspectives, self-assessed
communication skills, openness to learning, and behavioral change. Only 4 of 43
questions on the survey instrument yielded statistically significant shifts in opinions,
which highlights the challenges of changing deeply held beliefs through one brief
intervention.
Acknowledging the limitations of this experiment, the second half of this thesis
is a landscape analysis of civil discourse interventions at peer institutions UNC Chapel
Hill, California State University, Chico, and James Madison University. These
institutions provide promising models for integrating civil discourse into various
mandatory first-year experiences such as orientation or general education classes.
Finally, this thesis proposes specific policy recommendations for the University of
Delaware that highlight the need for a scalable, multi-touchpoint, embedded approach
for all students. The role of higher education is not just to teach academic content but
to also prepare students to be active and engaged democratic citizens.ente
Structural analysis of bijels stabilized by magnetically responsive ellipsoidal particles
No full textThis article may be downloaded for personal use only. Any other use requires prior permission of the author and AIP Publishing. This article appeared in
Nikhil Karthikeyan, Ulf D. Schiller; Structural analysis of bijels stabilized by magnetically responsive ellipsoidal particles. Physics of Fluids 1 August 2025; 37 (8): 083303. https://doi.org/10.1063/5.0268127 and may be found at https://doi.org/10.1063/5.0268127
© 2025 Author(s). Published under an exclusive license by AIP Publishing.
This article will be embargoed until 08/12/2026Bicontinuous interfacially jammed emulsion gels (bijels) offer a versatile platform for emulsion templating of functional porous materials including membranes, electrodes, and tissue-mimetic biomaterials. In many applications of such materials, the microstructure determines the properties and performance of devices. Characterization of the morphological structure of emulsion templates is thus an important step in developing fabrication methods for porous materials with tunable microstructure. We present a structural analysis of bijels stabilized by magnetic ellipsoidal particles. Using data from hybrid Lattice Boltzmann-Molecular Dynamics simulations of a binary liquid with suspended magnetic ellipsoidal particles, we analyze the bond orientational order within the interfacial particle layer, the mean and Gaussian curvature of the interfaces, and the topological properties of the emulsion morphology. The results suggests that the particle packing at the interface is influenced by the local topology as characterized by the Gaussian curvature, and the global topological properties can be linked to domain coarsening mechanisms such as coalescence of domains and pinch-off of channels. By analyzing independent simulation runs with different initial conditions, we probe the statistical variations of different properties, including the channel size distribution and the average channel size. Our analysis provides a more detailed picture of the structural properties of bijels stabilized by magnetically responsive ellipsoids and can guide the optimization of interfacial particle packing and domain structure of particle-stabilized emulsion systems.We thank the organizers of the 33rd Conference on Discrete Simulation of Fluid Dynamics held over 9-12 July 2024 in Zurich, Switzerland, for creating the platform at which, and for bringing together the audience to which, this work was first presented.
This work was supported by the National Science Foundation under NSF Award Nos. DMR-2414458 and OIA-2346036. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect those of the National Science Foundation. This work used NCSA Delta at the National Center for Supercomputing Applications through Allocation PHY220131 from the Advanced Cyberinfrastructure Coordination Ecosystem: Services & Support (ACCESS) program, which was supported by National Science Foundation Grant Nos. 2138259, 2138286, 2138307, 2137603, and 2138296
IDENTIFYING THE SOURCES OF DISSOLVED INORGANIC CARBON (DIC) AND CARBON ISOTOPES (δ13C-DIC) IN THE ROOSEVELT INLET (LEWES, DE, USA)
No full textThis study investigates biogeochemical variability in Roosevelt Inlet (Lewes, DE), where mixing between four water sources—Broadkill River (freshwater), Canary Creek (marsh), Lewes-Rehoboth Canal (brackish), and Delaware Bay (seawater)—shapes marine carbonate chemistry. The goal of this work is to use discrete endmember tide cycle sampling with continuous data logging to identify contributions of dissolved inorganic carbon (DIC) and δ¹³C-DIC from marshes and Delaware Bay to the Broadkill Estuary. Previous studies of DIC in coastal systems have rarely incorporated δ¹³C data, thereby limiting insight into carbon source attribution. To address this, we conducted five seasonal samplings at each water source where we combined discrete endmember high/low tide and high-resolution tide cycle sampling surveys on separate days and weeks within the same months. Measured pH, pCO₂, total alkalinity (TA), DIC, and δ¹³C-DIC identified physical and biological controls on marine carbonate chemistry. Isotopic signatures revealed the influence of organic matter decomposition, with δ¹³C values consistent with inputs from Spartina alterniflora (–13‰) and freshwater endmembers (–10‰). A two-endmember mixing model demonstrated seasonal effects on DIC and δ¹³C, where August and October values were driven by marsh-derived aerobic respiration rather than sulfate reduction. August and October low tides were dominated by marshland, producing high pCO₂ and low pH and dissolved oxygen (DO) values. December and February were influenced by the Delaware Bay endmember (–10‰), and reflected carbonate precipitation and cumulative aerobic processes, whereas April conditions indicated enhanced photosynthesis. Here, the mixing model effectively traced organic carbon sources and may be applicable to similar estuarine systems.National Science Foundation (NSF EPSCoR Project WiCCED, Grant No. 1757353) and the State of Delaware to the University of Delaware awarded to Wei-Jun Cai, NOAA ECOA MARACOOS 21-26 CEOE Cai (SMSP372289) and Dr. Wei Jun Cai’s start-up funding (SMSP175182)