19304 research outputs found
Sort by
Dual Energy Contrast Enhanced Mammography Using a Photon Counting Spectral Detector
No full textX-ray imaging is a crucial tool for material identification in a variety of fields including medical imaging, biology, security, and geoscience. Recent advancements in detector fabrication and technology have enabled creation of X-ray photon counting detectors with energy-resolving capabilities. The additional spectral information captured by these detectors has spurred development of more advanced material identification methods. One such method is dual-energy contrast-enhanced mammography, which is an important diagnostic tool for diagnosing lesions in dense breasts. Conventional digital mammography is an important tool for diagnosing cancerous lesions in breast tissue, but for dense breasts, healthy tissue signal can obscure cancerous lesions. Dual-energy contrast-enhanced mammography exploits a sharp rise in energy dependent attenuation of an injected contrast agent to provide additional lesion contrast. Using photon counting detectors for this task is ideal, as these detectors can simultaneously acquire the required energy dependent data. Despite their advantages, photon counting detectors suffer from spectral distortions which degrade image quality and quantitative accuracy. X-ray fluorescence of high-Z sensor material, Compton scatter within the detector, and charge sharing due to small pixel size all distort the measured spectra. In addition, beam hardening through the sample and object scatter from the sample further distort measurement. These issues must be compensated for to enable accurate dual-energy contrast-enhanced mammography measurements. In this research, a single material calibration method will be utilized to correct for the previously mentioned distortions. Each type of spectral distortion mentioned will be compensated for individually and in combination. This method will be applied to dual-energy contrast-enhanced mammography using a photon counting detector to improve contrast of an injected contrast agent
Modeling and Analysis of an AlN Piezoelectric Micro-swimmer with Integrated Gold Electrodes
No full textMicroscale robotics is expanding the possibilities for medicine, particularly in targeted drug delivery and in vivo diagnostics. Inspired by the natural propulsion mechanisms of microorganisms and recent advances in dislocation-driven growth shaping AlN's anisotropic morphology, this study explores a 2D model of an AlN piezoelectric micro-swimmer with integrated ultra-thin gold electrodes (0.045 μm) for controlled actuation and adaptability. The 2.89 μm AlN layer, combining stiffness and flexibility, harnesses piezoelectric effects to convert electrical energy into mechanical deformation. This deformation drives electric-induced kinematics, which, coupled with a closed-loop feedback control system, enables the micro-swimmer to demonstrate periodic motion, with potential for non-reciprocal actuation and navigation in low Reynolds number environments. The micro-swimmer's design features a multi-electrode configuration encapsulating the piezoelectric layer. Phase-shifted electric potentials and polarity differences across multiple electrodes produce alternating deformations. These dynamics are explored within a closed microchannel with pressure point constraints, as well as under a parabolic inlet velocity profile, revealing the swimmer's ability to resist flow-induced forces and maintain controlled motion. This study is simulated in a 2D environment due to computational limitations. To approximate and visualize potential 3D dynamics, an out-of-plane component was purposely selected. The study explores how parameters such as time-switch modulation, voltage control, and feedback mechanisms influence displacement and propulsion behavior. While achieving significant forward motion remains challenging, the swimmer exhibits controlled motion with potential for non-reciprocal actuation, contributing valuable insights into AlN micro-swimmer dynamics. These findings pave the way for future 3D simulations, such as a microhelix design, aimed at exploring AlN's robust piezoelectric characteristics, including the strong directional coupling along the z-axis, and efficient energy conversion, which are critical for effective propulsion. The insights gained from this study provide a foundation for subsequent investigations into the potential of AlN micro-swimmers for controlled navigation in microfluidic environments
Advanced Deep Learning Frameworks for Pollution Modeling: Applications in Numerical Solving, Model Emulation, and Uncertainty-Aware Air Quality Forecasting
No full textThis dissertation advances deep learning frameworks in atmospheric pollution modeling, presenting innovative approaches to numerical solving, model emulation, and uncertainty-aware forecasting. In Chapter 1, we conduct a comprehensive comparative analysis between standard finite difference methods and a deep learning-based solver for solving the 2D unsteady advection-diffusion equation. The DNN solver reduces mean and maximum absolute errors by up to 2 orders of magnitude while exhibiting enhanced stability in coarse spatial-temporal domains. The solver demonstrates consistent interpolation accuracy in resampled spatial and temporal domains magnified up to 5 and 16 times respectively, establishing its superiority over traditional methods. In Chapter 2, we develop a U-Net based deep learning emulator of the Community Multiscale Air Quality (CMAQ) model for surface NO2 simulations across the Contiguous United States. The emulator achieves a correlation coefficient (R) of 0.949 and Index of Agreement (IOA) of 0.974 when trained on 2011 and 2014 NEI data and evaluated on 2017 NEI data. The semi-normalized sensitivity of NO2 concentrations to NOx emissions aligns with CMAQ Decoupled Direct Method sensitivities. Diurnal analysis across 15 major urban environments reveals slight over- and underestimations of morning and evening peaks. The emulator significantly accelerates simulations by over 600 times using CPU and GPU compared to CMAQ. In Chapter 3, we implement a Temporal Convolutional Neural Network (TCNN) for bias-correcting CMAQ's 72-hour ozone forecasts over South Korea, incorporating uncertainty quantification. The TCNN improves R by 40.3%, IOA by 22.4%, and reduces Root Mean Square Error by 48.2% for day-1 forecasts, maintaining superior performance through day-3, while demonstrating robust spatial generalization. Shapley Additive Explanations analysis identifies past observed ozone as the primary predictor, with temperature's impact varying seasonally from first-ranked in winter to sixth in summer. Between two calibrated uncertainty quantification methods, Monte Carlo Dropout (MCD) outperforms Multi-Model Ensemble with a Spread-Skill Reliability of 0.23 ppb. The calibrated MCD effectively captures temporal uncertainty patterns with high correlation (R = 0.89) between error and uncertainty estimates across the 72-hour forecast. This research presents significant advancements in computational efficiency and accuracy for atmospheric pollution modeling, from numerical solutions to operational forecasting with uncertainty quantification
Alexithymia Moderates the Association for Racial Trauma and Negative Emotionality
No full textRecent research has revealed that experiences of racial discrimination produce symptoms of trauma and posttraumatic stress disorder. Available research shows that racial trauma is associated with adverse mental health outcomes for Black Americans, including negative emotionality (i.e., depression and anxiety). In order to inform interventions for racial trauma, it is essential to understand the mechanisms associated with the relationship between racial trauma and negative emotionality. One potential mechanism is alexithymia, or one’s capacity to identify and describe emotions. The purpose of this study is to examine alexithymia as a possible buffer in the association of racial trauma and negative emotionality among Black adults. There is little research regarding the Black American experience with alexithymia and negative emotionality. To date, there has been no literature examining the relationship between racial trauma, negative emotionality, and alexithymia. Therefore, the current study aims to explore these understudied areas. The participants of this study included Black American adults who indicated at least one experience with racial trauma. The current study is a secondary analysis of data that originated from a larger project that examined racial trauma and resilience in Black American adults. All analyses were conducted using IBM SPSS. Regression analyses were conducted via PROCESS Macro. For all models, racial trauma was evaluated as the predictor model, and alexithymia was evaluated as the moderator variable. Two separate analyses were conducted to examine the negative emotionality outcome: a) OASIS for anxiety and b) BDI for depression. Age and gender were included as covariates for all models. As predicted, results indicated that a) racial trauma and negative emotionality were significantly associated, and b) alexithymia functioned as a moderator between the association of racial trauma and negative emotionality. These findings provide insight into the importance of emotional expression in the context of race-based traumatic events
Lightning Electromagnetic Pulse Functions and Protecting the Artemis Spacecraft
No full textLightning produces some of the most powerful electric currents found on Earth, which are usually measured near the bottom or base of the lightning channel. However, these lightning “channel-base currents” also produce Electromagnetic Pulses (EMPs) which are powerful enough to be measured from the other side of the world. In this thesis, we introduce new channel-base current functions to represent lightning waveforms measured near the Artemis spacecraft, and we demonstrate that our functions can meet standard “Component A” specifications more closely than contemporary standards. We work with transcendental equations to peak-correct or “normalize” these functions, then solve for parameters graphically. To calculate the electromagnetic fields from lightning, we derive integrals from Lorenz and Maxwell’s equations, and we evaluate these integrals utilizing the lightning Transmission Line (TL) model, introduced by Uman and McLain in 1969. We assume the lightning return stroke rises at the speed of light to obtain perhaps some of the first analytical solutions to the TL model, using Euler substitution, hyperbolic substitution, integration by parts, and image theory to account for reflections. We also introduce a new polynomial substitution method as a possible alternative to Euler substitution. Finally, we introduce two new lightning models: the first resembles the Modified Transmission Line with Exponential decay (MTLE) model, while the second is likely the first model to have a variable return stroke speed. We additionally present several methods for protecting sensitive assets such as the Artemis spacecraft from lightning, both on the ground and during flight. We construct Computational Electromagnetics (CEM) models of Artemis and Launch Pad 39B using the Finite Element Method (FEM) in the time domain. We model a complex network of hanging catenary cables, each displaced by gravity and tension from other cables, and we encode them into cubic Bézier curves. We model the true shape of a lightning channel based on photographs from different vantage points, then simplify this geometry using an antenna array and a radiation pattern. We utilize Three-Dimensional Elevation Program (3DEP) data to construct the ground terrain surrounding our model. Finally, we compare our analytical and computational results to measurements
Unlocking the Power of Conversation
No full textThe ability to engage in meaningful and successful conversations impacts nearly every facet of our lives. It is through conversation that we build relationships, collaborate on work projects, share stories, teach children, communicate needs, and participate in community functions. But conversation is anything but simple. Conversational success involves much more than the words we say or the content of our message. During conversation, we participate in a rhythmic and dynamic interplay, continually altering our behaviors and coordinating these actions with one another. In this talk, I will discuss my research focused on how people coordinate their communicative behaviors with one another to improve conversational outcomes. I will then focus on the differences between the coordinative patterns of autistic and non-autistic individuals and how these differences can impact mixed- neurotype conversations. Finally, I will conclude with a discussion on what we are doing to try to create opportunities for better conversational experiences for autistic and non-autistic people alike
Using Self-Determination Theory to Understand Mental Health Risks Among Latina Americans
No full textDepression, anxiety, and stress symptoms among Latina American women have risen significantly in recent years despite strong familial-oriented systems theorized to protect members. The myriad of negative, positive, and null effects on mental health outcomes within this literature emphasizes the need for grounded motivational theories that can clarify cultural risk and protective effects on mental health. This study proposed that discrepancies between self-endorsed and perceived familial-endorsed marianismo beliefs would be associated with depression, anxiety, and stress symptomology, in part because these discrepancies undermine basic psychological needs (BPN) for autonomy and relatedness. I utilized the self-determination theory framework to enhance our understanding of marianismo endorsement, clarifying that frustration of autonomy and relatedness are key mediators between cultural value discrepancies and depression, anxiety, and stress within the Latina American population. This internal conflict can be understood by incorporating BPN frustration, which I hypothesized to mediate the association between marianismo belief discrepancy and emotional disorder symptoms. Three operationalizations of self and familial endorsed marianismo discrepancy tested whether these discrepancies would be associated with frustration of autonomy and relatedness, which in turn would be associated with depression, anxiety, and stress symptoms in Latina-American women. Support emerged for two of these operationalizations of belief discrepancies across two samples (Pilotn = 264; Primaryn = 343) spanning community and undergraduate college samples. For example, family-endorsement of marianismo beliefs, controlling for self-endorsement, was significantly associated with need frustration and, in turn, psychological symptoms. This pattern was not significantly altered when college attendance, generation status, and country of origin were included as potential main effects and moderators. These results emphasize the utility of examining familial cultural beliefs along with BPN to better understand associations with mental health among this population
Single Molecule Förster Resonance Energy Transfer and Super Resolution Force Spectroscopy Study of Elongation Factor G Evolution, Conformation, and Function
No full textThe ribosome is a ribonucleoprotein complex consisting of two subunits. During protein synthesis the ribosome decodes mRNA facilitated by a translocase enzyme, elongation factor G (EF-G). EF-G induces the advancement of the mRNA reading frame by 3 nucleotides (nt) and tRNA translocation via the hydrolysis of one GTP molecule. The mechanism of this reaction remains largely unknown. Cryo-EM studies indicated that the switch I region of EF-G undergoes major conformational changes following GTP hydrolysis, which may drive tRNA translocation. Multiple sequence alignment (MSA) of switch I identified a highly conserved threonine (Thr48) in prokaryotes, which may be homologous with Thr56 of eukaryotes. Two mutations were prepared: a phosphomimic (T48E) and a nonpolar substitution (T48V). Biological assays indicated both inhibited tRNA translocation. Single molecule Förster resonance energy transfer (smFRET) experiments determined this inhibition stems from the mutants becoming trapped in a ‘super’-compact conformation on the ribosome. The Thr48 mutations were modeled using AlphaFold predictions and MSAs. This revealed the mutations shifted the conformational balance to two previously reported EF-G conformations, con1 and con2. Predicted ancestral analogues were generated using phylogenetic tree construction. Two analogues resembling con1 and con2 were expressed. Both analogues exhibited minimal GTPase activity. Translocation fidelity was measured using super resolution force spectroscopy (SURFS), which found that both maintained the proper 3-nt reading frame. Structural studies have shown Q508 and H584 in domain IV of EF-G play important roles in translocation as they interact with codon-anticodon minihelix formed between mRNA and the A-site tRNA. We investigated EF-G mediated power stroke and translocation fidelity via mutagenesis of these two residues. Five substitution mutations were expressed (H584K, H584E, H584Q, Q508K, and Q508E). Biological assays have shown that all mutants retain uncompromised GTPase activity. Translocase activity was measured using SURFS. While Q508K maintained the proper 3-nt reading frame and produced normal power-stroke force, H584K induced a ‘-1’-nt frameshift and adversely affected power-stroke generation. smFRET experiments also indicated both mutations inhibit EF-G’s ability to enter a compact conformation when not bound to the ribosomal complex. Taken together, these studies provide us with new knowledge of GTPase coupled translocation and power-stroke force
Food Insecurity and Health Goals Among Young Adults Experiencing Homelessness
No full textBackground: Sources estimate that one third to 70% of YEH experience food insecurity in the US. Despite the evidence that food insecurity is connected to physical and mental health issues, few studies describe the need among this population. Objective: The purpose of this study is to describe food insecurity and health promotion needs among a sample of YEH. Design: Data was collected from YEH participating in a randomized controlled trial (n=450) testing the efficacy of an HIV prevention intervention. Methods: YEH between the ages of 16 and 25 were recruited from shelters and drop-in centers in a large urban area in the US. Baseline data was collected and participants answered questions related to food insecurity and health goals. Descriptive statistics were used to describe demographics, food insecurity, and health goals. Results: The mean age of participants was 21.1 years. Gender composition was mostly male (50%) and female (44%), and the majority of the sample was Black (62%). Over half of the respondents answered affirmatively for each food insecurity item, ranging from 50.9% to 63.1%. The highest reported health goal was feeling less stress (49.8%), with second and third highest being more physically active (46.9%) and eating more fruits and vegetables (42.4%). Conclusion and Implications for dietetics practice: Providers serving YEH in shelters and drop-in centers have an opportunity to reach YEH with programs to increase food access and health promotion strategies. Funding Source: National Institute of Nursing Research (NINR). [This project was completed with contributions from Diane Santa Maria from The UTHealth Science Center at Houston - Cizik School of Nursing.]Health and Human Performance, Department ofHonors Colleg
Generative Adversarial and Transformer Network Synergy for Robust Intrusion Detection in IoT Environments
No full textIntrusion detection in the Internet of Things (IoT) environments is increasingly critical due to the rapid proliferation of connected devices and the growing sophistication of cyber threats. Traditional detection methods often fall short in identifying multi-class attacks, particularly in the presence of high-dimensional and imbalanced IoT traffic. To address these challenges, this paper proposes a novel hybrid intrusion detection framework that integrates transformer networks with generative adversarial networks (GANs), aiming to enhance both detection accuracy and robustness. In the proposed architecture, the transformer component effectively models temporal and contextual dependencies within traffic sequences, while the GAN component generates synthetic data to improve feature diversity and mitigate class imbalance. Additionally, an improved non-dominated sorting biogeography-based optimization (INSBBO) algorithm is employed to fine-tune the hyper-parameters of the hybrid model, further enhancing learning stability and detection performance. The model is trained and evaluated on the CIC-IoT-2023 and TON_IoT dataset, which contains a diverse range of real-world IoT traffic and attack scenarios. Experimental results show that our hybrid framework consistently outperforms baseline methods, in both binary and multi-class intrusion detection tasks. The transformer-GAN achieves a multi-class classification accuracy of 99.67%, with an F1-score of 99.61%, and an area under the curve (AUC) of 99.80% in the CIC-IoT-2023 dataset, and achieves 98.84% accuracy, 98.79% F1-score, and 99.12% AUC on the TON_IoT dataset. The superiority of the proposed model was further validated through statistically significant t-test results, lower execution time compared to baselines, and minimal standard deviation across runs, indicating both efficiency and stability. The proposed framework offers a promising approach for enhancing the security and resilience of next-generation IoT systems