Published ETD Collection

Permanent URI for this collection


Recent Submissions

Now showing 1 - 20 of 5660
  • Item
    (2022-12-13) Cheng, Kuan; Ardebili, Haleh; Karim, Alamgir; Ghasemi, Hadi; Bao, Jiming; Ryou, Jae-Hyun
    Conventional techniques to harvest and store energy are challenged by the ever-increasing demand for versatile forms of electrical energy caused by the rapid expansion of the Internet of Things (IoTs). As emergent solutions, flexible triboelectric nanogenerators (TENGs) and lithium-ion batteries (LIBs) have been invented and extensively studied in recent years. Different TENGs are fabricated to scavenge mechanical energy from most natural sources and human motions, making them portable solutions to energy generation on-demand. On the other hand, rechargeable LIBs play critical roles in the evolution of energy over 50 years, owing to its abilities to store massive amount of energy, lay the foundation for portable smart devices, and make possible a fossil fuel-free world. At the beginning of this dissertation, latest efforts that incorporating low-dimension carbon materials with TENG systems will be systematically reviewed. Carbon materials, including graphene and carbon nanotube, can bring many synergistic properties to TENGs, such as output enhancement and multifunctionality. They are poised to further the reach of TENG applications and make a positive impact on common issues related to TENG technology. The second section is to present a robust route to fabricate flexible TENGs with multifunctionality by nano-patterning thermoplastic polyurethane (TPU) thin films. Topographically optimized TENGs could promote higher power generation while preventing biofilm formation without using any chemical additives. Analysis of pattern amplitude and wavelength correlation to output power is uniquely provided for a deeper understanding of how patterned TENGs enable peak performance. The last part of this work presents the fabrication and characterization of lithium-ion batteries based on solid-state polymer electrolytes. Efforts made to substitute conventional liquid electrolyte and plastic separators make a great accomplishment on mechanical properties and safety aspects of LIBs. Fluoroethylene carbonate (FEC) has been proved as an effective electrolyte additive, which helps building LIB systems with ultra-high capacity and low self-discharge. Comprehensive electrochemical properties along with thermal properties of LIBs will be closely scrutinized in this work.
  • Item
    A Three-Part Study Investigating Parent Perceptions of Control Regarding Their Child’s Autism Spectrum Disorder
    (2023-08-17) McNeel, Morgan Margaret; Mire, Sarah; Keller-Margulis, Milena A.; Kim, Hanjoe; Storch, Eric
    Background: While children with autism spectrum disorder (ASD) experience symptoms that affect them as individuals, the impact of ASD extends beyond the diagnosed child. Despite strengths, parents of children with ASD face unique challenges and experience higher levels of stress and depression. It is critical to understand and address parent-specific factors that may negatively affect families and, ultimately, children. Parent cognitions, such as perceptions and beliefs, influence coping and have implications for important health-related behaviors, such as those related to intervention needs, decision-making, and processes. More specifically, better understanding parent perceptions regarding control, a construct related to parenting self-efficacy, may aid researchers and healthcare providers in improving support for families of autistic children. Purpose: This three-part study investigated: (1) parent, child, and family variables as predictors of parents’ perceived control; (2) how parents’ perceived control may change over time, and possible relationships between those changes and changes in other child, parent, and/or family factors; and (3) relationship between parents’ perceived control over ASD symptoms and their treatment decision making, when also considering other parent, child, and family characteristics. Methods/Results: Study 1 used data from 362 parents of children with ASD. Multiple linear regression revealed that parents’ perceptions of more/higher control over their children’s symptoms was predicted by lower Parental Distress, child age, and overall ASD symptom severity, as well as higher Positive Coping Skills. Study 2 examined parent perceptions of control related to their child’s ASD for a subset of 16 parents across two time points alongside other individual, parent, and family characteristics. Change in parents’ perceptions of control over symptoms was examined at the item-and scale-level, and findings included that perceptions of control among parents in this small sample were generally high and relatively stable over time, though the sample size limits interpretation and generalizability of these findings; the clinical significance of the change was also considered. Study 3 used a subset of data from 327 parents from study 1. A series of binary logistic regression analyses identified several predictors of ever using specific autism-focused interventions among families in this sample. Parent perceptions of effects of the child’s diagnosis, perceived controllability of symptoms, and perceived severity of core ASD symptoms were the most common significant predictors of families endorsing ever having used certain popular interventions. Conclusion: Outcomes of these interrelated studies have both research and practice implications for better understanding and supporting parents and families of children with ASD. The studies highlight the importance of further exploring parent perceptions regarding control and self-efficacy when working with this population.
  • Item
    The Impact of Teachers’ Words: A Phenomenological Study of Teachers’ Perceptions of Their Talk Used with Students
    (2023-08-16) Thornell, Sydney Taylor; Hutchison, Laveria; Cooper, Jane; Reis, Nancy; Thomas, Dustine
    Background: School districts and other educational systems have extensive options of programs to choose to implement for various types of professional and student learning. These well-intentioned and often research-supported programs and learning systems aim to improve education for students. A problem is that once a program is adopted, emphasis is often placed on its potential and routine classroom usage, rather than on how it is being implemented and if it is effective for students. Instruction on how teachers speak and present content is missing, leaving educators, especially novice teachers, unable to attain the full potential of the adopted program. Successful implementation of programs and curricula is directly impacted by teachers’ talk. Purpose: The purpose of this study was to explore teachers’ talk in their classroom settings, with students, and their perceptions of it after participating in professional development on the topic of teachers’ use of classroom talk. Teacher talk in this study is defined as the words educators say to students and the tone of voice in which they say them. Professional development is focused on warm, respectful, and student-centered talk moves. The following research question guided this study: What are the perceptions of both novice and experienced teachers toward their use of talk used with students before and after professional development? Methods: This study employed a qualitative phenomenology approach to collect and analyze data from eight participants. Four of the participants in the study were teachers at the study site with less than one year of teaching experience in the profession. The other four participants in the study were experienced teachers at the study site but with less than one year of teaching experience in the school district. Data sources included semi structured one-on-one interviews with participants designed by the researcher and reviewed by an educational expert, participants’ reflections on classroom talk and interactions with students, a contemporaneous researcher field journal used to document the researchers’ experience and observations throughout the study, and audio recorded semi-structured group conversations with participants. Both thematic and discourse analysis were used to make sense of the data collected. Member-checking was also utilized to ensure accuracy in representing participants’ perceptions of teacher talk. Results: The findings from this study revealed four themes: Professional development on teacher talk improved both novice and experienced teachers’ awareness of their teacher talk used with students; Both novice and experienced teachers recognized areas of growth in their own teacher talk after professional development on the topic; New-to-district and novice teachers benefited from the campus-designed induction program where professional development was presented, and lastly; New-to-district and novice teachers shared the feeling that professional expectations, including teacher talk, were different for them when compared to their more established colleagues. Conclusion: The perceptions of both novice and experienced teachers in this study have shown that the phenomenon of professional development on the topic of teacher talk has increased self-awareness and sparked a change in the way that teachers talk with students.
  • Item
    Atmospheric Composition Changes During Droughts in the Continental U.S.
    (2023-04-22) Li, Wei; Wang, Yuxuan; Jiang, Xun; Flynn, James H., III; Li, Liming
    The abnormal meteorological conditions under droughts can impose large changes in atmospheric compositions. In this dissertation, we quantified these changes using long-term atmospheric composition observations over the continental U.S. during summertime. Specifically, we revealed the spatial east-west variation in ozone (O3) response to drought: higher O3 enhancement in the southeast U.S. (SEUS) and no significant change or even a decrease in the west. We attributed this spatial discrepancy to O3 chemistry caused by the opposite response of isoprene: a 37% decrease in isoprene under exceptional drought in California in contrast to a 41% increase in Georgia. The enhanced isoprene in the SEUS also contributes to the 24% higher value of organic aerosol (OA), which can be largely attributed to the increase of isoprene epoxydiols derived secondary organic aerosol (IEPOX SOA) with a high dependence on sulfate. The elevated OA in the Pacific Northwest under droughts is caused by increasingly higher wildfire emissions. We evaluated the chemical transport model GEOS-Chem regarding its capability in capturing the observed drought-air pollution relationships. The model under- and overestimates the drought-induced O3 and OA changes in the SEUS, respectively, which can be partly caused by the overprediction of biogenic isoprene emissions. A satellite-derived drought stress factor by minimizing the model-to-observed bias of formaldehyde column to temperature sensitivity was implemented in GEOS-Chem. The resulted reduction in isoprene emissions can lower the OA positive bias by 7%-12% and improve the O3 enhancement by 1-3 ppb over low-NOx regions. We also found a decrease of 11% in surface fine dust over the SEUS under droughts in contrast to the expected increase in other regions. Through the teleconnection to the negative North Atlantic Oscillation, a lower-than-normal and more northeastward displacement of the Bermuda High is present during SEUS droughts, resulting in less dust being transported into the SEUS. The enhanced precipitation in the Sahel associated with the northward shift of the Intertropical Convergence Zone also leads to lower dust emissions therein. The GEOS-Chem model can capture the weakened African dust transport and reproduce the reduced dust in the SEUS while misses the enhanced dust in the western areas.
  • Item
    Methodologies for Evaluating and Interpreting Neural Code Intelligence Models
    (2023-04-24) Rabin, Md Rafiqul Islam; Alipour, M. Amin; Gnawali, Omprakash; Kakadiaris, Ioannis A.; Hellendoorn, Vincent J.
    Deep neural models are increasingly being used for various code intelligence tasks, such as code summarization, auto code generation, and software bug detection. Researchers commonly utilize these models to solve different downstream tasks for improving developer productivity and code quality. Despite the continuing development of code intelligence models, it remains largely unclear how reliable these models are in real-world scenarios. This issue is further complicated by the fact that these models are opaque black-box and depend on noise-prone data sources for learning. Therefore, to reliably adapt such models, researchers often need to reason about their underlying behaviors and the factors that affect them. However, our understanding of how generalizable these models are on unseen data and what relevant features they learn for making predictions is largely unknown. A lack of knowledge in these areas may exaggerate the learning behaviors of models and can lead to reckless deployment in safety-critical applications. Moreover, state-of-the-art approaches are typically specific to a particular set of architectures and require access to the model's parameters, which hinders their reliable adoption by most researchers. To address these challenges, we propose a set of model-agnostic methodologies that inspect models by analyzing inputs and observing outputs without accessing the model's parameters. The overarching goal is to enhance our understanding of the model's inference by exploring its learning behaviors in terms of generalizability and interpretability. Specifically, we assess the ability of a model to generalize its performance with respect to noise-inducing memorization and semantic-preserving transformation. Additionally, we identify critical features from input programs for interpreting the predictions of a model through prediction-preserving reduction. Our results indicate that neural code intelligence models are prone to memorizing noisy data due to their excessive parameters, are often vulnerable to very small semantic changes, and typically rely on a few syntactic features for making their predictions; thus, models usually suffer from poor generalization performance in unseen scenarios. These observations could assist researchers in better understanding the underlying behavior of these models and prompt them to focus their efforts on devising new techniques to alleviate the shortcomings of existing models.
  • Item
    Machine Learning Estimation of Daily Surface Concentrations of PM2.5, NO2, and MDA8 Ozone at High Spatiotemporal Resolutions
    (2023-03-06) Ghahremanloo, Masoud; Choi, Yunsoo; Jiang, Xun; Rappenglueck, Bernhard; Lefer, Barry L.
    High concentrations of pollutants in the atmosphere endanger public health and negatively impact other domains. Although surface measurements of pollutants at ground stations are quite reliable, they still suffer from the low spatial coverage due to the limited number of ground stations. Such limitations call for the development of accurate approaches to estimating surface concentrations of pollutants, particularly in regions with no monitoring stations. This thesis proposes machine learning (ML) and deep learning (DL) techniques to estimate daily surface concentrations of PM2.5, NO2, and daily maximum 8-h average (MDA8) ozone. The first task focuses on using random forest (RF) to estimate daily surface concentrations of PM2.5 at 1-km spatial resolution in the 2014-2018 period over Texas to obtain a correlation coefficient (R) of 0.83-0.90 and a mean absolute bias (MAB) of 1.47-1.77 µg/m3. Our results also show the high capability of RF compared to the commonly used models for estimating PM2.5 concentrations. The second task focuses on developing the PCNN-DNN , a novel two-step DL model, to estimate daily surface NO2 concentrations over the contiguous United States (CONUS) from 2005 to 2019. To the best of our knowledge, the PCNN-DNN is the most accurate model in the globe to estimate surface NO2 levels, with an R of 0.975 to 0.978 and an MAB of 0.99 ppb to 1.38 ppb. Moreover, the PCNN-DNN model generates estimated NO2 grids without any missing values, improving the quality of various applications such as public health studies. The third task is to develop a DL approach to accurately estimate surface MDA8 ozone and examines the spatial contribution of several factors on ozone levels over the CONUS in 2019. The model obtains an R of 0.95 and an MAB of 2.79 ppb, highlighting the promising performance of the Deep-CNN at estimating surface MDA8 ozone. We also use Shapley additive explanations (SHAP) to generate, for the first time, a spatial feature contribution map (SFCM) for ozone, the results of which confirm an advanced ability of Deep-CNN to accurately capture the relationships between ozone and most predictor variables.
  • Item
    3D facial modeling with geometric wrinkles from images
    (2023-04-27) Deng, Qixin; Deng, Zhigang; Pavlidis, Ioannis T.; Chen, Guoning; Mayerich, David
    Realistic 3D facial modeling and reconstruction have been increasingly used in many graphics, animation, and virtual reality applications. Currently many existing face models are not able to present rich details while deforming, which means lack of wrinkles while face shows different expressions. Also, to create a realistic face model for an individual is also needs complex setup and sophisticated works from experienced artists. The goal of this dissertation is to achieve an end-to-end system to augment coarse-scale 3D face models, and to reconstruct realistic face from in-the-wild images. I propose an end-to-end method to automatically augment coarse-scale 3D faces with synthesized fine scale geometric wrinkles. I define the wrinkle as the displacement value along the vertex normal direction, and save it as displacement map. The distribution of wrinkles has some spatial characteristics, and deep convolutional neural network (DCNN) is pretty good at learning spacial information across image-format data. I labeled the wrinkle data with its identity and expression vectors. By formulating the wrinkle generation problem as a supervised generation task, I implicitly model the continuous space of face wrinkles via a compact generative model, such that plausible face wrinkles can be generated through effective sampling and interpolation in the space. Then I introduce a complete pipeline to transfer the synthesized wrinkles between faces with different shapes and topologies. The method can augment an exist 3D face model with more fine-scale details, but to create a realistic human face model is not yet solved. Properly modeling complex lighting effects in reality, including specular lighting, shadows, and occlusions, from a single in-the-wild face image is still considered as a widely open research challenge. To reconstruct an realistic face model from an unconstrained image, I propose a CNN based framework to regress the face model from a single image in the wild. I designed novel hybrid loss functions to disentangle face shape identities, expressions, poses, albedos, and lighting. The outputted face model includes dense 3D shape, head pose, expression, diffuse albedo, specular albedo, and the corresponding lighting conditions.
  • Item
    Automatic Content Understanding for Safe and Positive Media Experiences
    (2023-05-01) Zhang, Yigeng; Solorio, Thamar; Chen, Guoning; Eick, Christoph F.; Gonzalez, Fabio A.
    In this dissertation, we introduce the task of automatic content understanding for media. Digital media products, such as movies, songs, and books, serve not only as sources of entertainment but also as vehicles for knowledge dissemination. While ubiquitous in people's daily lives, media content may not always be suitable for users of all ages. For instance, movies with sexually suggestive language or violent scenes might be inappropriate for children. Policymakers have developed systems like the Motion Picture Association (MPA) film rating system for movies and the Parental Advisory Label (PAL) for music to classify age suitability for media products. Likewise, service providers such as streaming websites invest effort in offering suitability suggestions to their customers. However, current rating systems depend on expert classification, resulting in an expensive and inefficient process. Furthermore, these ratings provide only general age-based suitability labels, offering limited information for users seeking deeper content insights. While media can have an educational impact, leveraging it as a resource for teaching morals and building character, especially for children, can be challenging. For instance, parents may need an appropriate story to teach their children about honesty. Manually sifting through every product to identify those with the desired educational value is impractical, even for experts in the field. We address these shortcomings by automating the media content understanding process through various machine learning (ML)-based formulations using natural language processing (NLP) techniques. We have two standing points in this research: One is protective action for safe experiences: first, we investigate rating movie severity in different age-restricted aspects (such as violence and sex) to provide perceptible level information as a compliment for the general suitability category. Then we expand the scope from movies to music products to assess the song lyrics for not only the risky aspects but also the positive messages. The other is proactive action for positive experiences: we go one step further to study how and to what extent the NLP model can interpret themes and educational values from the literature. Then we study how a story reaches a positive or negative narrative outcome with the moral it intends to convey. All of our efforts will go towards one goal: providing comprehensive information for media products. In line with these research topics, we first formulate media content understanding tasks as machine learning problems and then create benchmark datasets for movies, music, and literature. Based on these benchmarks, we successfully designed, implemented, evaluated, and analyzed novel methods to address the research problems. Our proposed techniques demonstrate superior performance compared to existing methods and pave the way for future methodological exploration. These research outcomes will significantly contribute to the goal of providing people with safe and positive media experiences.
  • Item
    Small Spacecraft Design & Machine Learning-based Approaches To Lunar Robotics Navigation
    (2023-05-08) Tanaka, Toshiki; Malki, Heidar A.; Becker, Aaron T.; Song, Gangbing; Cescon, Marzia; Provence, Robert S.
    Since human exploration of the Moon in the 1960s, the lunar community has benefited from a series of successful missions, including flybys, orbiters, landers (crewed and robotic), rovers, and impactors. The next generation of lunar exploration will include a cis-lunar station, crewed missions, and in-situ resource utilization (ISRU)-based missions that will generate significant amounts of data to help answer questions about how the Moon formed and evolved, what its surface processes and resources are, and the nature of the chemical composition of its surface and deep interior. Complete utilization of the currently available technologies is vital to effectively plan and execute future missions. This can be facilitated by two key technologies: small satellites and machine learning (ML). Nowadays, satellite technologies have progressed to the point where off-the-shelf components can be purchased for small-satellite missions, greatly reducing the time and cost needed to prepare a new mission. The rapid escalation of the production and launch of small satellites has revolutionized the space industry, proving that small satellites in constellations are more useful than fewer, larger ones for some scientific missions and radio relay missions on a large scale. ML and artificial intelligence also play an increasingly important role in aerospace applications, particularly for automated systems, including space robotics guidance, navigation, and control. This dissertation aims to demonstrate three potential components that small satellites and ML could help accelerate in view of future exploration of the Moon and other planetary bodies. The discussion is divided into three topics: 1) renewal of lunar navigation systems with small spacecraft, 2) a machine learning-based approach to lunar hopper control, and 3) a machine learning-based approach to small rover path planning. In the first topic, a new triangulation theory that enables the creation of lunar global navigation satellite systems with just two small satellites is introduced. In the second topic, a new ML-based methodology for lunar hopper obstacle avoidance, descent, and landing is presented. In the third topic, a new ML-based global path planning methodology for small lunar rovers is proposed.
  • Item
    Borehole Seismic Methods: Estimating Anisotropy And The Distributed Acoustic Sensing (DAS) To Geophone Transformation
    (2023-05-07) Sayed, Ali Yawar; Stewart, Robert R.; Chesnokov, Evgeni M.; Zheng, Yingcai; Kumar, Dhananjay
    Several vertical seismic profile (VSP) methods are addressed with the aim of improving accuracy of the results or extracting new insights from the data. An effective horizontal transverse isotropy (HTI) framework was built to relate Thomsen’s parameters to fracture density and fracture fluid. Direct shear energy present on the horizontal components of vertical-vibrator zero-offset VSP data on two field data examples was used to characterize the fast-shear azimuth along well depth. Parametric wavefield decomposition was used for the strong anisotropy case and the rotation-correlation method was adapted for the weak anisotropy case. The results correlated with independent log measurements and studies. Synthetic signatures of azimuthal VSP data generated from weakly, moderately, and heavily fractured model show that fracture response of dry fractures is stronger than that of fluid-filled fractures. Similar signatures using a dipping interface within isotropic and fractured models show that isotropic subsurface structures can produce an apparent fracture response and they can distort the true fracture response. A structure-consistent orientation workflow was developed to correct for the structural effect and uncover the true fracture response. Beyond conventional VSPs, a transformation was developed to convert distributed acoustic sensing (DAS) measurements to conventional velocity/acceleration measurements. A theoretical framework was developed to show that DAS data are inherently filtered in spatial frequency and amplitude. Synthetic and field data examples are analyzed to show that DAS measurements have a detrimental effect on traveltime picking, Q-estimation, and imaging. These effects are negated by the DAS-to-velocity transform. The methods developed in this dissertation are being actively used to deliver commercial products in the field.
  • Item
    Biomarker Panel Array Systems for Disease Detection
    (2023-05-11) Tang, Chenling; Wu, Tianfu; Mohan, Chandra; Ning, Jing; Shevkoplyas, Sergey S.; Li, Zhengwei
    Lupus nephritis (LN) is a devastating chronic kidney disease (CKD) caused by Systemic lupus erythematosus (SLE), an autoimmune disease that involves a loss of immune tolerance to endogenous materials and causes inflammatory responses and multiple organ damage. Currently, invasive renal biopsy as the gold standard diagnosis for LN may cause kidney injury, especially for multiple biopsy tests. Moreover, the histopathological morphology classification system doesn’t stratify the heterogenous nature of lupus patients. Fortunately, serum and urine biomarkers could potentially serve as a surrogate of disease activity, and a biomarker panel composed of multiple biomarkers could largely improve the diagnostic or prognostic sensitivity and specificity. By applying a LN biomarker panel on a multiplexable protein microarray platform, we aim to build a robust, highly sensitive, multiplexed, and high-throughput point-of-care system for lupus nephritis. This research covers four sections: (1) Discovery of novel circulating immune complexes (ICx) in the serum of lupus nephritis as potential biomarkers. Immunoproteomics-based discovery studies combined with Bioinformatics-assisted selection have enabled us to identify circulating immune complexes as potential biomarkers of LN. (2) Discovery of novel serum biomarkers that have diagnostic or predictive value in lupus nephritis. A novel serum biomarker VSIG4 has been discovered using proteomics and further validated as a promising novel serum biomarker of lupus nephritis and renal pathology activity. (3) Development of a Point-of-Care serum Biomarker-Panel Mini-Array (BPMA) system for the diagnosis, disease monitoring, and flare prediction of lupus nephritis; BPMA-S6 has been developed as a novel promising POC device for LN diagnosis, disease monitoring and flare prediction of LN (4) Discovery of novel urine biomarkers and biomarker-panel that have diagnostic or predictive values or disease monitoring capability for lupus nephritis. Collectively, this study has led to the discovery of novel biomarkers, the identification of a novel 6-plex biomarker panel, and the development of a novel biomarker panel detection system for autoimmune kidney disease. These findings hold great promise for the next generation of home care and community medicine.
  • Item
    Predicting Reservoir Quality in Organic-rich Bakken Shales, North Dakota, Using 3D Seismic and Petrophysical Analysis
    (2023-05-02) Paris Castellano, Andrea Gloreinaldy; Stewart, Robert R.; Castagna, John P.; Li, Aibing; Verm, Richard
    A petrophysical analysis, using well logs along with empirical and theoretical models, is undertaken to understand and characterize organic-rich prospective intervals within the Bakken Formation in North Dakota. A more heuristic and data-driven approach for unconventional petrophysics modeling proposes a model based on a priori partitioning of the pore system and fluids enabling reasonable prediction of the volumetrics of the shale rock using a minimum suite of logs. The total porosity and water saturation curves estimated using the revised model agree favorably with effective porosity and water saturation core measurements in both the shale members of the Bakken Formation. This suggests that kerogen volume controls the total porosity of shale reservoirs, and the total porosity models should honor the influence of both the inorganic and organic phases of the rock with a more accurate fluid distribution. We include calculations and mapping of interval properties from log data and integrate the seismic with seismic AVA analysis and anisotropic fluid substitutions to explain the observed low velocity ratios within the thin Bakken shales as being due to hydrocarbon saturation and kerogen presence. The modeling of the Bakken shale rock frame needs the inclusion of both kerogen and pyrite content for it to be in better agreement with expectations from Greenberg-Castagna (GC-92) empirical shale trend. Integrating well-log analysis, rock physics, seismic modeling, multi-linear regressions, and supervised machine-learning predictions of organic-richness, total porosity, and water saturation contributes to identifying the distribution of promising reservoir quality areas within the Bakken shale members.
  • Item
    (2023-04-27) Panthi, Bikash; Pinsky, Lawrence S.; Ma, Jingfei; Hwang, Ken-Pin; Ratti, Claudia; Timmins, Anthony
    Neoadjuvant systemic therapy (NAST) is administered prior to surgery to reduce tumor burden in patients with triple-negative breast cancer (TNBC); however, about half of TNBC patients do not respond to NAST and develop a distant spread within five years. Early assessment of NAST response for triple-negative breast cancer (TNBC) is critical for patient care, both to avoid toxicities from ineffective treatments and to provide novel targeted therapies to the non-responders. The purpose of this study was to explore and then validate functional magnetic resonance imaging (MRI) techniques as imaging biomarkers capable of early assessment of NAST response in TNBC. First, we investigated the potential of tumoral and peritumoral radiomic features extracted from dynamic contrast-enhanced (DCE)-MRI and diffusion-weighted imaging (DWI) to predict response to neoadjuvant systemic therapy (NAST) in TNBC. We explored and validated these features on 163 patients (training set = 109, testing set = 54) at three different time points: before the initiation of NAST (BL), after two cycles of NAST (C2), and after four cycles of NAST (C4). We identified 152 radiomic features with an area under the curve (AUC) ≥ 0.70 for both the testing and training cohorts. The relative changes in the 1st and 5th percentile between BL and C4 had AUCs ≥ 0.80 for both training and testing sets. Secondly, we developed multivariate radiomic models based on multiparametric MRI images and investigated them to predict NAST response. Forty-nine multiparametric MRI-based models had AUCs > 0.75. The top-performing radiomic model used 35 radiomic features and had AUCs of 0.91 and 0.80 for training and testing sets, respectively. Finally, we assessed functional tumor volumes (FTVs) from DCE-MRI as predictors of response to TNBC. FTVs were measured for 100 patients at BL, C2, and C4 using optimized percentage enhancement (PE) and signal enhancement ratio (SER) thresholds. The maximum AUC for predicting treatment response using FTV was 0.84 at C4, followed by FTV at C2 (AUC = 0.82). Our study demonstrated the potential of radiomic analysis of multiparametric MRI and FTV as non-invasive biomarkers for early prediction of treatment response in TNBC patients.
  • Item
    Extending the OpenSHMEM Communication Model Into the Exascale Era
    (2023-05-07) Welch, Donald Aaron; Subhlok, Jaspal; Wu, Panruo; Gnawali, Omprakash; Chapman, Barbara M.; Hernandez, Oscar R.
    OpenSHMEM is a partitioned global address space (PGAS) programming model that is used in many application domains for the simplicity of its memory abstractions and proximity to the low level remote direct memory access (RDMA) network performance features it provides. However, its origin was as a thin interface to specific low-latency hardware, and as such aged along with the system architecture that gave rise to it and is no longer as naturally suited to express the features of modern and emerging architectural trends. Of particular concern are difficulties with scaling small and irregular remote memory accesses on modern interconnects along with the increasingly diverse memory types they access. This work investigates what kinds of extensions to its design may be required to better adapt to the diversity of future computing systems whilst remaining as close as possible to the traditional OpenSHMEM model and retaining its low overhead. To that end, we will first look at the construction of a highly performant reference implementation capable of directly leveraging latent capabilities of network technology via a performance portable low-level communication middleware in order to drive public interest and development. Then, we will look at a series of extensions to the OpenSHMEM model aimed at increasing its flexibility and maximizing its performance, notably to increase memory and interprocess locality for exploiting resource affinity, and message aggregation abilities to achieve optimal use of network resources. We will additionally show a proof of concept for how the OpenSHMEM model can be mapped to a high productivity language like Python. Finally, we will present a tool intended to use high level patterns for identifying regions of applications that could benefit from these new extensions as well as analyze the resulting application performance behaviors.
  • Item
    Fabrication and Characterization of Low Dimensional Metal Oxides and Sulfides for Solar Energy Conversion
    (2023-05-11) Rana, Dhan; Varghese, Oomman K.; Chen, Shuo; Stokes, Donna W.; Robles Hernandez, Francisco C.; Hosur, Pavan
    Increasing energy needs and the environmental damage caused by carbon dioxide (CO2) accumulation in the atmosphere due to fossil fuel burning are the two most pressing issues in today’s world. A fast transition to renewable sources-based energy system could address these issues. Solar photovoltaics (PV) and solar photoelectrochemical (PEC) water splitting are two promising renewable energy conversion technologies. Nevertheless, the PV and PEC technologies must offer affordable power and fuels, respectively, for making an impact in the energy market. The primary objective of this dissertation work was to develop and characterize novel nanostructured semiconductors for efficient solar energy conversion. Considering the potential for efficient solar energy conversion, material abundance, environmental compatibility and economic viability, tungsten oxide and copper tin sulfide were selected for the investigation. We employed a colloidal synthesis method to obtain CTS sol. Electrochemical anodization, a low cost and scalable method, was used to fabricate nanostructured tungsten oxide. A major invention that emerged from the dissertation work was the growth of tungsten oxide nanotubes using anodic oxidation. Except for an unconfirmed work, no study had ever shown the fabrication of WO3 nanotubes of length in the micrometer scale. We identified the fabrication conditions favorable for growing ordered nanoporous and nanotube array films of WO3 in a wide thickness range. We carried out a comprehensive investigation of the effects of additives, oxidizers and solvent composition in the electrolyte and other synthesis conditions on the growth of anodic nanostructures of WO3. The photoanodes fabricated using this material for PEC water splitting showed ~75% improvement in the photocurrent compared to the highest reported for a W/WO3 photoanode. CTS thin films were fabricated using a newly developed sol synthesis method. The sol was highly air stable. Performance of CTS solar cells employing WO3 films as window layers was compared with sol TiO2 based CTS cells. Although the PV performance was not highly impressive, the study showed that CTS and WO3 nanotube could be promising for future photovoltaics. The dissertation discusses the details of the development processes of these new materials and their properties relevant to devices for solar energy conversion.
  • Item
    QCD Equilibrium and Dynamical Properties from Holographic Black Holes
    (2023-05-05) Grefa, Joaquin; Ratti, Claudia; Bellwied, Rene; Hosur, Pavan; Labate, Demetrio
    Strongly interacting matter undergoes a crossover phase transition at high temperatures and zero net-baryon density. A key question in Quantum Chromodynamics (QCD) is whether a dense quark-gluon system exhibits critical phenomena when breaking the balance between quarks and antiquarks. Lattice QCD studies suggest that a critical point can only emerge in the baryon-dense domain, which is challenging to describe through ab initio calculations. In this dissertation, I employ a bottom-up Einstein-Maxwell-dilaton (EMD) holographic model, which reproduces first principle lattice QCD thermodynamics at zero and small density and reflects the near-perfect fluidity of the quark-gluon plasma (QGP), to describe the hot and dense QGP at finite temperature and density. I significantly extend the baryon density coverage of the equation of state for hot and dense quark-gluon matter thanks to new numerical techniques to map holographic black hole solutions to the QCD phase diagram. This allows us to locate the predicted critical point and the first-order phase transition line over a wide region of the phase diagram. Comparisons with the most recent lattice results for the QCD thermodynamics are also presented. The EMD model is also employed to determine several transport coefficients of the hot and baryon-rich quark-gluon plasma at the crossover, the critical point and across the first-order phase transition line. These transport coefficients include the shear and bulk viscosities, baryon and thermal conductivities, baryon diffusion, jet quenching parameter, heavy-quark drag force, and Langevin diffusion coefficients. Finally, Bayesian inference methods are applied within the EMD model to forecast the QCD equation of state throughout the phase diagram. We numerically derive the posterior probability distribution for holographic model parameters based on lattice data at zero chemical potential and extract their most probable values. Enhanced calculations allow us to sample a large number of data fits using Monte Carlo techniques, providing an estimate for the critical point's location and the associated statistical error bands. This research offers insights into the dense quark-gluon plasma, its transport coefficients, and the critical point location, which can be probed in heavy ion collision experiments.
  • Item
    Understanding The Visual And Auditory Defect In Ush2a Mouse Model
    (2023-05-08) Crane, Ryan; Naash, Muna I.; Al-Ubaidi, Muayyad R.; Groves, Andrew K.; Yang, Jun; Porter, Jason; Romero-Ortega, Mario I.
    Usher syndrome (USH) is the most common form of dual deafness and irreversible vision loss found in patients worldwide. USH2 is the most prevalently occurring sub type, accounting for ~50 to 75% of USH clinical cases. Patients with USH2 suffer from congenital hearing loss and progressive vision loss beginning from adolescence. Mutations in USH2A (usherin) account for ~80% of USH2 patients; making it the most common genetically mutated gene among USH patients. Usherin has been detected in the photoreceptors of the retina and in the developing inner ear cell stereocilia. The USH2A gene is very large and attempting gene therapy with conventional viral delivery is not easily attainable due to limited payload capacity of commonly used Adeno-associated viruses (AAV) ( <4.7 kbp.) To understand the mechanism underlying hearing and visual impairments, a knock in (KI) mouse model (Ush2adelG/delG) was developed using one of the most prevalent human mutations of USH2A, 2299delG. While other models for Ush2a exist, no other KI model has been able to successfully mimic the genetic mutations found in humans. This model will be valuable to investigate possible delivery methods for USH2A. Retinal phenotype in the KI model was found to follow similar progression as patients, with a notable gradual loss of vison that became apparent at older ages, along with concomitant photoreceptor degeneration. More in-depth analysis found that this was due to the mislocalization of the KI mutant protein and a subsequent mislocalization of its other USH2 interacting partners. Along with patients exhibiting phenotypes associated with homozygous mutations of USH2A, there is also a prevalence of cases involving heterozygous mutations of USH2A in combination with mutations in other USH or non-USH genes that results in varying degrees of phenotype. Backcrossing of Ush2adelG/delG with three knockouts of photoreceptor-specific proteins, rhodopsin (Rho-/-), ABCA4-/-, and ROM1-/-, resulted in double heterozygous of digenic mutants with a mixture of retinal phenotype. The combination of heterozygous USH2A with rhodopsin (Rho+/-/Ush2adelG/+) led to a surprising protective effect of retinal phenotype. We next evaluated the cochlear phenotype of the KI model, showing congenital hearing loss in the lower frequency range, particularly at 8 and 11 kHz, in the auditory brainstem response (ABR) tone tests. Associated stereocilia of the inner hair cells in the apical portion of the cochlea were correspondingly found to be disorganized. Like the retina, the Ush2adelG/delG mutant protein was found to be mislocalized while the associated proteins were found to be properly localized to the stereocilia of the hair cells. Following the analysis and verification of the KI model as a viable model for therapeutic testing, the next step was to develop a therapeutic approach for both the retina and cochlea. With one major challenge of therapy for USH2 being its very large gene, development of an alternative, non-viral delivery, method is needed and an example of such an approach is by encapsulating the therapeutic gene using hyaluronic acid nanospheres (NSs). Delivery approaches for the eye, specifically targeting the retina, using NSs in conjunction with a small molecule, sulfotyrosine, have shown great promise. Preliminary studies using NSs filled with plasmid DNA with GFP expression cassette under the control of ubiquitous promotion (chicken beta actin with CMV enhance, CAG) led to noticeable GFP expression in retinal cells following intravitreal injections. No GFP was detected with the plasmid DNA alone. Along with the retina, delivery to the inner ear cells of the cochlea was also a goal. Preliminary injections of the NSs in P0-P1 round window injections into the cochlea showed no functional differences between the injected and non-injected (contralateral) ears, indicating no major toxic effect caused by the NSs. Future work will look more in depth at the possibility of delivering native USH2A genes into the retina and cochlea as a potential therapeutic options.
  • Item
    Investigating Gaze Orientation and Spatial Localization in Strabismus
    (2023-05-15) Karsolia, Apoorva; Das, Vallabh E.; Stevenson, Scott B.; Nurminen, Lauri; Tamber-Rosenau, Benjamin J.
    Purpose: Disruption of binocular vision during the critical period of development results in strabismus in 3-5% of the population. The visual system adapts to this decorrelation with the help of suppressive mechanisms that influence eye choice behavior. The overall goal of this research was to identify visual and non-visual factors that may impact gaze orientation and localization behavior in strabismus. These studies shed light on the mechanisms underlying binocular vision and spatial localization and provide insights into the temporal dynamics of visual suppression and its impact on eye-choice behavior. Methods: In Aim 1, development of horizontal and vertical ocular alignment was assessed in six prism-reared infant monkeys using Hirschberg photographic methods. In Aim 2, ten human subjects with normal ocular alignment localized briefly presented targets, presented to same or alternate eyes under dichoptic conditions, in an unreferenced environment. In Aim 3, eye movements were recorded in two adult exotropic monkeys while performing memory saccade tasks with variable delays to assess influence on persistence of visual information on fixation-preference. Results: Aim 1: Monkeys reared with prisms during infancy developed strabismus as early as 3 weeks of age (~3 months in humans) suggesting influence of both visual and non-visual mechanisms in development of normal alignment. Aim 2: Under conditions of binocular competition (dichoptic viewing), human subjects were unable to compensate for their inherent phoria and made greater errors as compared to same-eye viewing condition. Aim 3: Fixation preference behavior was observed in adult prism-reared monkeys during memory-guided saccades, similar to patterns observed during visually-guided saccades. Memory delays up to 800msec did not alter fixation preference behavior. Conclusion: Prism-reared monkeys mimic strabismus in humans and are a useful model to study its behavioral and neurophysiological implications including influence of oculomotor proprioception. Binocular dissociation in absence of visual cues, leads to inaccurate localization even in normal ocular alignment, indicating that extra-retinal eye position feedback in the form of oculomotor proprioception may be imprecise or derived from the wrong eye. Visual suppression in strabismus leads to long-lasting adaptations that influence eye choice behavior beyond the stimulus presentation time (800ms), as indicated by fixation patterns of localization.
  • Item
    (2023-05-11) Wang, Yifei; Majd, Sheereen; Abidian, Mohammad Reza; Wu, Tianfu; Chow, Diana Shu-Lian; Du, Guangwei
    Nanoliposomes are one of the most commonly used delivery nanocarriers due to their biocompatibility, biodegradability, and low toxicity. To achieve high levels of cellular uptake, liposomes often require large doses of fusion-promoting molecules or targeting ligands that can lead to undesired side effects, including toxicity and immunogenicity. To address this challenge, this project aims to utilize the biological process of membrane phase-separation to design and develop liposomes that can offer highly efficient cellular internalization with minimal toxicity. First part of this dissertation combines experimental and computational tools to investigate the phase behavior of multi-component lipid membranes. The experimental studies focused on studying phase-separation on micron-sized liposomes of various compositions using fluorescence microscopy. In collaboration with mathematicians, two continuum phase-field models were then developed to simulate the phase-separation examined in experiments. Great agreement between experiments and simulations validated the computational models and demonstrated their potential use for the design of phase-separating and patchy liposomes. Second part of this dissertation explores the use of phase-separation to create highly fusogenic liposomal nanocarriers with minimal toxicity. The impact of charged lipids on membrane’s phase behavior was first investigated in multi-component micron-sized liposomes. The findings of this work were then applied towards designing fusogenitic liposomes with cationic patches that showed enhanced fusogenicity compared to their homogenous counterparts. This work demonstrated that phase-separation can be applied to enhance the performance of cationic delivery liposomes. The last part of this dissertation seeks to use phase-separation to enhance cellular uptake of ligand-conjugated liposomes. Focusing on biotin-streptavidin binding, as a model system, biotinylated liposomes were designed to respond to acidic pH in tumor environment to undergo phase-separation and present their ligands in highly-dense patches. We are further investigating the application in cell studies. Together, this study provides an insight into the use of phase-separation to control the functionality of lipid membranes and it hence, offers new possibilities to overcome the shortcomings of current liposomal nanocarriers.
  • Item
    Thermodynamics of the QCD Equation of State Near Deconfinement and the Critical Point
    (2023-04-24) Nava, Angel; Ratti, Claudia; Bellwied, Rene; Quaini, Annalisa; Koerner, Lisa W.
    The strong nuclear force is one of the four fundamental forces of our universe, and it is responsible for the stability of the atomic nucleus, through its binding of protons and neutrons, as well as the existence of all the hadrons which have been discovered. The quantum field theory which describes this force is quantum chromodynamics (QCD), and the exploration of this theory—through various experiments, models, and theoretical tools—has unveiled a phase diagram which we are still working to understand. There are two main features of this phase diagram which are central to my dissertation. First, is the deconfinement phase transition, where hadrons are melted into the quark-gluon plasma (QGP) via a smooth crossover at low chemical potentials. This exotic phase of matter was created microseconds after the Big Bang and it eventually cooled down into the hadronic matter which surrounds us today. Second, is the search for a critical point (CP) on this phase diagram, as the smooth crossover is expected to become a first-order phase transition at higher chemical potentials. In this dissertation I will outline my work of creating partial pressures which enable us to see the contributions from different hadron families (distinguished for the first time by all three conserved charges BSQ) to the full pressure given by lattice QCD, the state-of-the-art simulations by which QCD can be solved numerically. Each partial pressure exhibits non-monotonic behavior, the signature of the onset of deconfinement for that respective hadron family. In addition, I present an improved open-source code that produces families of equations of state (EoSs) which include a CP at a user-defined location along the transition line. The updated EoS adds constraints which are relevant to Heavy-Ion Collisions (HICs) where physicists routinely create and study the QGP. An EoS is a required input for hydrodynamic simulations which describe these collisions, and it can help guide experimentalists in their search for its location on the QCD phase diagram, since it enables us to study signatures of the CP. I will then use this critical effect to understand how the mapping of the 3D Ising model CP onto the QCD phase diagram affects the net-proton kurtosis.