Washington University Medical Center
Washington University St. Louis: Open ScholarshipNot a member yet
26344 research outputs found
Sort by
Ultra-Sensitive Whispering Gallery Mode Resonators for Photoacoustic Bio-Sensing and Imaging
No full textWhispering gallery mode (WGM) micro resonators are robust optical devices that have high sensitivity, high-quality factors, small mode volume, and strong light-matter interac- tion. This report delves into the extensive applications of WGM microresonators, with a specific focus on their pivotal role in photoacoustic biosensing and imaging. Even though the microresonators themselves can detect nanometer-sized particles utilizing a shift in res- onance, they have to rely on evanescent field interactions with molecules on the surface, while molecules outside the range remain undetected. This report introduces a novel ap- proach, proposing the generation of photoacoustic signals with a pulsed laser and the detec- tion of photoacoustic signals through the microresonator, to enable subsequent imaging or biomolecule detection. Early detection of pathogens, particularly bacteria, is essential for accurately determining the stage of wound infection. Different types of bacterial colonies dominate at different stages of infection due to changes in the wound environment—such as pH, oxygen levels, and the immune response. Identifying the dominant bacteria at an early stage can guide appropriate treatment decisions and help avoid unnecessary use of broad-spectrum antibiotics, which contributes to the growing global issue of antibiotic resistance. This is especially critical in developing regions like Asia and Africa, where access to advanced medical diagnostics is often limited. While traditional methods like PCR or genome sequencing provide accurate bacte- rial identification, they are typically bulky, expensive, and time-consuming. Our approach offers a faster, more accessible alternative. In our current work, three kind of wound bacteria samples are introduced into the microbubble resonator sensor and selectively excited by a pulsed laser at a wavelength specific to the bacterias. Our technique enables selective detec- tion and classification of bacterias by exploiting their unique optical absorption properties. While our current work focuses on controlled flow conditions, it can be potentially miniatur- ized for smart wound dressings, wearable patches and AI assisted monitoring system. The high sensitivity of whispering-mode-resonator makes it an excellent candidate for the high throughput bacteria detection across an extended sensing volume laying the foundation for future non-invasive, contactless wound infection bacteria sensing. In the latter section of the report, we detail the utilization of a packaged microbubble res- onator for imaging, leveraging the acquisition of photoacoustic signals. Even though we have been using tapered optical fiber for coupling light into microresonators, its’ instability and fragility pose limitations when applied in practical applications. The packaging of WGM sensors for practical using MY-132-A, a material distinguished by its low refractive index compared to alternatives. This advancement enables the effective application of the pack- aged device in photoacoustic imaging techniques, offering a comprehensive understanding of the operational capabilities of WGM microresonators in real-world scenarios
Firing the Canon: Queer Representation in the Magical Girl Genre
Full text linkThis research delves into the impact queer representation in the shoujo (Japanese for young girl) genre has on audiences. More specifically, the “Magical Girl” (a subgenre of shoujo) genre’s main focus is about ordinary girls whose lives are drastically changed when they are given magical powers. One of the major plot points of magical girl media is how these girls find balance with their ordinary lives with their magical alter egos, and their attempts to keep these lives separate. Much of this media follows a small group of girls, who tend to build deeply emotionally intimate relationships as they share this secret identity with one another. In some instances, these relationships can blossom into romantic ones, while others are left up to interpretation. In this research, I have found LGBTQ+ representation can also be vital to queer communities, as many have experienced discrimination due to their gender or sexual identities. Due to this prejudice, many fans have turned to the magical girl genre, as these shows and characters can bring comfort reminicisent of their childhood. Additionally, animated and illustrated characters can have a longer shelf life than living actors, as they are scarcely under scandolous fire, allowing audiences to be lifelong fans. The building of these fan audiences can have created what is considered fan culture, or fandom. Fan culture can have agreat impact on how these shows and relationships are interpreted, doing reparative work through fanart, fanfiction, and cosplay. With the creation of magical girl content whether it be from fans or creators, I argue that the genre can be reflective of the queer experience
Shell Quilt: Piecing Memory through Quilting, Collage, and Printmaking
Full text linkShell Quilt is a mixed media quilt and accompanying book which collect an amalgamation of found, hand dyed, and printed fabric pieces. This project serves as a visual tether and inquiry into my late grandmother Sylvia’s life and a way of piecing together memories of my time with her. As a conceptual frame for my thesis project, Shell Quilt, I have looked to the process of femmage, a term initially coined by Miriam Schapiro and Melissa Meyer, which I interpret as a portmanteau of feminism and collage. Through the lens of femmage, I explore the processes of collage, quilting, and print mediums and processes, in the work of the Folly Cove Designers, as well as artists like Margaret Kilgallen, Hương Ngô, Kiki Smith, and Faith Ringgold
Undefinitive
Full text linkMy thesis project, Undefinitive, examines the cyclical process of destruction, transformation, and rebirth through installation, ceramics, and text. With a deep unease toward my own gendered body, I investigate how vulnerability, both physical and psychological, can become a site of resistance and empowerment. Drawing from feminist theory, Chinese mythology, and stream-of-consciousness narratives, my work explores the monstrous and the fragmented as inseparable states of one’s becoming. Through references to body art, speculative fiction, and queer theory, I reimagine femininity and queerness as forces capable of dismantaling societal structure and forging new modes of existence.
Set in a post-apocalyptic world where consciousness inhabits non-human forms, Undefinitive tells a queer love story between a moth and an orchid at the dusk of the human era. Using text as the foundation for worldbuilding, I construct a fragmented installation where ceramic sculptures and video projections evoke fictional histories that resist linearity and definition. By inhabiting the in-between space of human and non-human, body and mind, I explore how gender, love, and identity might persist beyond the collapse of traditional binaries. In doing so, I seek to envision a new ontology rooted in fragmentation, transformation, and queer affinity
Influence of Laser Power on Glass Formation in Cu64Zr36 Fabricated by Direct Laser Deposition
Full text linkMetallic glasses, metallic materials with a disordered, non-crystalline atomic structure, can be fabricated by rapidly quenching a specific alloy from the liquid state. One intriguing fabrication strategy is direct laser deposition (DLD), and additive manufacturing technique which involves using a laser to melt small volumes of material, which then cool rapidly once the laser is removed. The goal for this independent study is to find how the laser power during the first remelt affects the overall glass formation ability of Cu64Zr36 alloy fabricated from elemental powders. This was done by using a Laser Engineered Net Shaping machine (LENS) to melt and linearly deposit elemental powders onto a zirconium substrate. These samples were characterized by using optical microscopy and measuring glassy sections with ImageJ. By defining the glassy structure as the surface of the sample being smooth and continuous, and a glassy section being at least fifty percent glassy by line width, it was found a remelt power of 290 W gave the largest total percentage of metallic glass at 40.22%. The most consistent line was measured by looking at which sample had the longest glassy segment. This was at 260 W, with the largest segment measuring 1829.9 μm
Comparative Analysis of Intrinsic Reward-Based Reinforcement Learning Algorithms
Full text linkReinforcement learning agents often struggle in tasks with sparse or delayed rewards, since they receive little guidance about which actions to pursue. This thesis investigates how adding intrinsic rewards can help address that issue. We focus on three main methods: Count-Based bonuses, where states are hashed and infrequent states receive higher rewards; Random Network Distillation (RND), where a predictor network learns to match the output of a fixed random target; and the Intrinsic Curiosity Module (ICM), which uses an inverse and a forward model to highlight transitions the agent cannot yet predict.
We implement these approaches under a single Proximal Policy Optimization (PPO) framework and evaluate them on three environments: Cartpole, MiniGrid, and Breakout. The results show that each method excels under certain conditions. Count-Based exploration performs well when the state space can be discretized effectively, while RND and ICM scale better to complex or pixel-based domains. However, none of the methods is universally dominant. The thesis concludes that factors such as dimensionality, reward structure, and the ease of hashing or modeling states should guide the choice of exploration bonus in sparse-reward reinforcement learning
Reimagining the American Union
No full textThis book is a thought experiment. It invites you to imagine an America without state government. The hypothesis it tests is that, all things considered, the national interest of the United States would be better served by a two-layer (national and local government) unitary system than by the current three-layer (national, state, and local) federation. In that hypothetical system, the functions currently performed by state government would be redistributed among the national government, the local governments, and various inter-government partnerships. The country would cease to be a union of states. It would become a union of its people.
This exercise – one that the scholarly literature has not yet undertaken – requires a careful, objective weighing of both the benefits and the costs of state government in the United States. That evaluation encompasses, but goes beyond, the traditional debates over federalism. The question considered here is not merely whether the US can do without federalism, but whether it can do without state government entirely. Those are different inquiries. While it is impossible to have federalism without political subdivisions, it is very possible to have political subdivisions without federalism. Indeed, unitary systems in which the subunits are mere subordinates of the central government are commonplace in today’s world. In such a unitary system, it seems fair to ask how many levels of subordinate units are optimal. This book suggests that for the United States, the answer is one – local government
Applying Machine Learning to Developmental Psychopathology
No full textPsychopathology arises from a combination of many contributing factors, which can include genetic predispositions, family history, environmental stressors, developmental processes, and psychosocial dynamics. Each of these contributing factors has been researched for decades, and their influences on mental health have been well documented. In recent years, there has been a growing enthusiasm for integrating these risk factors using machine learning. However, there are a wide array of data-driven frameworks that have been implemented, and the relative strengthens of each approach is not clear. Through of series of four research studies, we examined each data-driven approach in terms of its potential to derive accurate predictions and enhance our understanding about the developmental origins of psychopathology. Our work identified critical methodological challenges when using machine learning to predict atypical aging (linked to multiple forms of psychopathology). Specifically, there were multiple sources of bias that hindered generalizability and no clear systematic ways of improving the utility of brain age gaps. These issues were not mitigated by multimodal imaging, as the only improvements were to predictive accuracy, which did not impact the utility of brain age gaps. Next, we provided novel insights about feature selection for deriving brain-based models of psychopathology, and further insight into how age-related differences should be incorporated into the model. Lastly, we implemented an RDoC approach towards building risk calculators of psychopathology, which uncovered key insights about the importance of sex-differences and social relationships for mental health. Mental health is a complex psychological construct, and deriving accurate predictive models will require greater integration across multi-level predictors and more customization to better account for societal differences and cultural norms regarding mental health
Computational Imaging Under Incomplete Information
No full textComputational imaging is a pivotal field that synergizes physical measurement principles with advanced algorithms to generate visual information. An important task in this field is solving imaging inverse problems that aim to reconstruct high-quality images from observed measurements. Model-based deep learning (MBDL) has emerged as a particularly powerful tool for tackling these inverse problems by integrating machine learning (ML)-driven priors with knowledge of the imaging physics. This dissertation focuses on the pervasive challenge of informational incompleteness in computational imaging, arising from various practical and physical limitations, that hinder the widespread adoption of ML-driven computational imaging algorithms in practice. This includes: (a) the incompleteness of datasets that lack ideal ground-truth references for training machine learning models (Part II); (b) incomplete knowledge of the imaging system\u27s physical forward model (Part III); and (c) the incompleteness of acquired data in multimodal imaging scenarios (Part IV). To address dataset incompleteness, this dissertation pioneers various self-supervised algorithmic frameworks. These include methods that leverage inter-measurement consistency for dynamic imaging (Phase2Phase, Chapter 3), employ deformation-compensated learning for motion-affected scenarios (DeCoLearn, Chapter 4), and establish provable techniques for training advanced implicit neural networks without reference ground truth (SelfDEQ, Chapter 5). To address incomplete forward model knowledge, various MBDL techniques are introduced for blind inverse problems that aim to estimate both images and forward model unknowns. These feature block-coordinate plug-and-play algorithms with convergence guarantees that integrate deep denoiser priors on all unknowns (BCPnP, Chapter 6), and self-supervised deep unrolling strategies for concurrent system calibration and image reconstruction in parallel MRI (SPICER, Chapter 7). For incompleteness in acquired multimodal data, unified diffusion models are proposed for diverse reconstruction and synthesis tasks from arbitrary combinations of available input modalities (Any2All, Chapter 8). Furthermore, this thesis demonstrates the broad applicability of MBDL through its successful validation in challenging biomedical applications (Part V), including deformable image registration (PIRATE, Chapter 9) and fast ptychographic image reconstruction (PtychoDV, Chapter 10). Collectively, this dissertation significantly enhances the capacity of computational imaging to function reliably, practically, and effectively under diverse conditions of informational incompleteness through innovations in algorithm design, theoretical understanding, and practical application, paving the way for more powerful imaging solutions
Through the Looking-Glass: Analytical Flexibility and Interpretive Consensus in Resting-State fMRI
No full textMost data analyses underscore an obvious but difficult question: what information do we lose in the process of extracting something we can understand? This question emerges in widespread obstacles in the analysis and interpretation of resting-state fMRI data. In neuroimaging, extensive post-processing of resting-state functional MRI (rfMRI) data is necessary for its application, especially when investigating neural mechanisms of mental illness. However, widespread variability in these post-processing pipelines has hindered both the reproducibility and accumulation of knowledge in this area of the field. A choice of dimension-reduction algorithm is one of the most variable and impactful ones made in an rfMRI post-processing pipeline, and the interpretive implications of this choice will be the primary focus of this defense. We study information consensus between dimension reductions by examining the variability they induce on the topology (or, more precisely, the persistent homology) of the Human Connectome Project. We also probe this example case for general insights into the topological stability of dimension reduction problems (of n D-dimensional points to n d-dimensional points) in the regime D ≫ d ≫ n, which is not mathematically well-characterized despite its practical prevalence and significance