22689 research outputs found
Sort by
AUTOMATION METHODS FOR INTERVENTIONAL AND DIAGNOSTIC WEARABLE ULTRASOUND
No full textUltrasound imaging stands as one of the cornerstones of modern diagnostic and interventional medicine due to its unique advantages of being real-time, inexpensive, and radiation-free. Ever since its inception in the 1950s, advancements in technology have dramatically reduced the size of ultrasound machines and increased their functionality as well as accessibility. Most recently, the advent of wearable ultrasound devices has pushed the boundaries even further: the ultrasound imaging capabilities have been seamlessly integrated into a compact, wearable form factor that aims to deliver continuous monitoring in diagnostic and interventional applications.
Different from the prior generations in the ultrasound machine evolution where an expert hand is always required to perform ultrasound exams, a pivotal element for the successful deployment of wearable ultrasound devices lies in their capability to autonomously perform imaging acquisition, coupled with automatic image analysis and visualization with minimal human intervention. Therefore, this dissertation focuses on the development of such enabling technologies during the implementation of a novel wearable ultrasound device for both diagnostic and interventional applications. Technologies presented in the dissertation can be categorized with the following goals:
- Simulation technologies for the development and validation of a new wearable mechatronic ultrasound scanner.
- Automatic image acquisition that aims to obtain patient-specific optimal data with the wearable scanner’s 2 degree-of-freedom volumetric imaging capability.
- Image analysis algorithms that process the acquired data and generate either a diagnostic result in a continuous monitoring setting or an information-condensed summary such as a reconstructed bone model during the interventional guidance.
- Integration and validation with various visualization techniques such as augmented reality in the interventional guidance setting.
\end{enumerate}
These technologies will be synergistically demonstrated in two representative clinical scenarios for the diagnostic and interventional aspects: (a) fetal ultrasound monitoring for diagnostic application, and (b) lumbar puncture guidance for the interventional application
IMMUNE DETERMINANTS OF HOST SURVIVAL FROM ALPHAVIRUS-INDUCED ENCEPHALOMYELITIS
No full textAlphaviruses are arthropod-borne viruses that can cause encephalomyelitis. Viral infections of the central nervous system (CNS) present a challenge for the immune response because cytolytic responses cannot be deployed due to the non-renewable nature of neurons. The overarching goal of this dissertation is to further define the mechanisms and requirements for successful recovery from viral infection of the CNS using a mouse model of infection with Sindbis virus (SINV). The first chapter aims to identify the mechanism by which IRF7 promotes survival from SINV infection. IRF7 is a transcription factor that is essential for the induction of IFNα, and Irf7-/- mice lack IFNα expression. Selective deletion of Irf7 in peripheral myeloid cells resulted in death – despite intact expression of IFNα – with increased inflammatory monocyte infiltration into the CNS. This suggests that pathology in full-body Irf7-/- mice is driven by IRF7-deficienct monocytes and indicates a type I IFN-independent role for IRF7 in regulating inflammation. The second chapter aims to characterize the role of IFNβ in recovery from SINV infection. Compared to WT mice, Ifnb-/- mice had increased viral titers, inflammation, and pathology. These changes were associated with decreased microgliosis, increased inflammatory chemokine expression, increased infiltration of monocytes, and decreased inflammation of NK cells, highlighting a role for IFNβ in regulation of innate immunity. Lastly, the third chapter aims to identify changes to peripheral immune cells associated with CNS-restricted viral infection. The adaptive immune response is initiated in the periphery in draining lymph nodes prior to its migration into the CNS, and blood is a compartment seldom studied in mice. Single-cell transcriptomics of peripheral blood mononuclear cells identified unique cell populations and transcriptional signatures associated with SINV infection of the CNS. Additionally, in the brain, the appearance of discrete lymphocyte populations occurred concomitantly with increases of those same populations in the blood, which identifies potential blood markers associated with viral encephalitis. Taken together, these data highlight multiple immunologic phenomena that are associated with recovery and disease severity following SINV infection, adding to the mechanistic understanding of pathogenesis and providing potential targets for the development of novel therapeutic strategies for viral encephalitis
Selective Synaptogenesis in the Development of Cell-Type-Specific Connectivity in the Cerebral Cortex
No full textThe cerebral cortex contains many neuronal cell types, which differ in morphology, electrophysiology, and functional roles in cortical circuits. These cell types exhibit selective synaptic connectivity, which may be regulated by axonal and dendritic morphology, neuronal activity, and molecular guidance mechanisms. However, how preferential synaptic connectivity emerges during the development of excitatory cortical neuron types, which represent over 80% of the neurons in the cortex, is still poorly understood.
I used the synaptic connections of excitatory layer 6 corticothalamic neurons (L6CThNs) in mouse somatosensory cortex, which are strongly biased toward connections with inhibitory parvalbumin-positive (PV) interneurons over other excitatory neuron types, to uncover developmental mechanisms for excitatory intracortical synaptic selectivity. Preferential synaptic connectivity may be generated through selective initial synaptogenesis, or by promiscuous synaptogenesis that is followed by selective elimination of off-target synapses. I showed that L6CThNs specifically form synapses onto PV interneurons during initial synaptogenesis in L6 and in the development of interlaminar inputs to L4. L6CThNs did not form detectable silent synapses in neonatal L6, further indicating that synaptogenesis occurs specifically. Activity-dependent plasticity mechanisms that act via calcium-permeable AMPA receptors, which are enriched in PV interneurons, may also affect selective synaptogenesis. However, I found that reducing the calcium permeability of AMPA receptors in PV interneurons did not alter connectivity from L6CThNs to PV interneurons. Axonal overlap with synaptic targets is necessary but not sufficient for synaptogenesis in development, and the temporal relationship between intracortical axon growth and synaptogenesis is not fully understood. I showed that L6CThN synaptogenesis occurs during the postnatal elaboration of L6CThN axons. L6CThN axon development occurs in two distinct phases, with a first phase of axon growth in infragranular layers followed by a delayed second phase of axon elaboration in L4. Thus, the synapses of L6CThNs in L6 and L4 develop during layer-specific phases of axon elaboration but exhibit selective synaptogenesis across both layers. This thesis demonstrates the selective formation of L6CThN outputs that regulate inhibition across layers during adult cortical function and provides insight into the mechanisms that govern excitatory synaptic targeting within the cerebral cortex
Alzheimer's Disease- and Aging-Mediated Blood-Brain Barrier Dysfunction in a Tissue-Engineered Microvascular Model
No full textAlzheimer’s disease is a disease of neurodegeneration and aging that affects millions of Americans, and is expected to impact millions more without further significant breakthroughs. Though years of preclinical and pharmaceutical investment have yielded some advances in treatment strategies, current approaches only modestly alter disease progression. A critical underlying difficulty in Alzheimer’s treatment is a lack of understanding of the etiology and progression of Alzheimer’s, especially given the complex interactions of many different molecular, cellular, and environmental cues that are correlated with phenotypic outcomes. An emerging focus in Alzheimer’s study is the role of the cerebrovasculature in the initiation, progression, and exacerbation of symptomatic disease. Disruption of the blood-brain barrier, which tightly controls any exchange between systemic circulation and brain tissue, has manifested in post-mortem and in vivo studies of late-stage Alzheimer’s disease as microbleeds, dysfunctional glucose transport, and impaired efflux of toxins; additional animal studies have indicated that some vascular dysfunction precedes neuronal degeneration in the progression of the disease. Thus, to understand the drivers and progression of Alzheimer’s disease in hopes of identifying therapeutic breakpoints, the role of blood-brain barrier dysfunction must be investigated. To do so, here we utilize a tissue-engineered model of the blood-brain barrier with high spatiotemporal resolution to assess its dysfunction under key categories of perturbation associated with Alzheimer’s disease. These perturbations will span extrinsic cues of oxidative stress (hydrogen peroxide exposure), the systemic influence of aged blood components (exposure to aged vs. young human serum), cell-intrinsic mutations associated with Alzheimer’s (APP(Swe) and PSEN1(M146V)), and select combinations thereof. This combinatorial approach allows for modular study of each contributor and their impact on transcriptome, proteome, and blood-brain barrier function; meaningful functional changes to highlight include the disruption of cellular patency under acute and chronic oxidative stress, increased transcellular transport with exposure to aged human serum in a tissue-agnostic microvessel precursor, and additive exacerbation of paracellular permeability, endothelial activation, and angiogenesis with the combination of intrinsic Alzheimer’s-related mutations and circulatory cues of aging
Solving a Stochastic Dynamical System with the Lattice Boltzmann Method
Get PDFThis thesis explores several numerical approaches to determining the state probability densities of a stochastic dynamical system, including an approach called the Lattice Boltzmann Method that is seldom used in this application. The stochastic differential equations describing a two-state stochastic dynamical system are rewritten into the Fokker-Planck equation that describes the time evolution of the state probability densities as a partial differential equation. The Fokker-Planck equation takes the same form as the advection-diffusion equation, which describes the evolution of densities of physical concentration, and can be solved using the Lattice Boltzmann Method. We have identified an extant Multiple Relaxation Time variant of the Lattice Boltzmann Method that shows promise to be used in the instances where a stochastic system contains states that do not have stochastic state derivatives.
Four stochastic systems are posed and integrated from initial conditions using four different approaches that include two Lattice Boltzmann Method variants, a Monte Carlo approach, and an analytical solution. The resulting probability density solutions from the possible methods are compared and the Lattice Boltzmann Method is shown to have practically low and continuous errors
DEFORMATION AND FLOW MODULATION OF FINITE SIZED BUBBLES IN INTENSE TURBULENCE
Get PDFTurbulence remains one of the most captivating phenomena in classical mechanics due to its complex nonlinear relationships spanning a wide range of scales. Introducing an immiscible second phase, such as bubbles, adds significant complications, including bubble induced agitation, modified power spectra, and energy and mass exchange through surface deformation. This thesis aims to advance the understanding of this classical multiphase flow problem by investigating how individual bubbles modulate turbulence in their vicinity and how turbulence affects bubble dynamics, particularly surface deformation.
The first part of this thesis focuses on turbulence modulation in the wake of deforming bubbles in homogeneous and isotropic turbulence, where the turbulence intensity is around 1. Due to the low void fraction, bulk turbulence statistics remain unaffected by the bubbles, and bubble-bubble interaction is negligible. Thus, the focus is on the effect of individual bubble wakes on local turbulence. To quantify the flow and bubble deformation, we measured the fluid flow in 3D with tracer particles using Lagrangian particle tracking and simultaneously reconstructed the 3D geometry of the bubbles using six high-speed cameras with the visual hull method. Unlike in a quiescent medium or weak turbulence, the wake does not have a persistent direction. The decorrelation time scales of slip velocity are roughly equal to the bubble-sized eddy turnover time, suggesting that local turbulence frequently alters the direction of the slip velocity. Therefore, to study local turbulence modulation, we employed an axisymmetric reference frame with the center of symmetry aligned with the direction of the slip velocity. This allows for averaging turbulence statistics near a bubble, exposing the wake signature in intense turbulence. The results suggest that local turbulence is augmented and strongly dependent on the bubble Reynolds number, the orientation of the bubble's major axis relative to the slip velocity, and the bubble aspect ratio.
The second part of this thesis examines bubble interfacial energy fluctuations driven by turbulence by tracking the velocity of the most stretched tip of the deformed bubble in 3D. The results show that the power spectrum based on the tip velocity exhibits scaling similar to that of the Lagrangian statistics of fluid elements but decays with a distinct timescale and magnitude modulated by the Weber number based on bubble-sized eddies. This indicates that the interfacial energy is primarily siphoned from eddies of similar sizes as the bubble. Additionally, the tip velocity appears more intermittent than the velocity increment at the bubble scale, emphasizing the key roles played by small-scale eddies in extreme events.
These findings provide a framework for understanding the energy transfer between individual deformable bubbles and multiscale eddies in intense turbulence, paving the way for more complicated studies such as large void fraction experiments where bulk turbulence modulation is evident
Interactive Evaluation of LLM-based Agents
Get PDFPersonality design plays an important role in chatbot development. From rule-based chatbots to LLM-based chatbots, evaluating the effectiveness of personality design has become more challenging due to the increasingly open-ended interactions. A recent popular approach uses self-report questionnaires to assess LLM-based chatbots' personality traits. However, such an approach has raised serious validity concerns: chatbot's “self-report” personality may not align with human perception based on their interaction. Can LLM-based chatbots “self-report” their personality? We created 500 chatbots with distinct personality designs and evaluated the validity of self-reported personality scales in LLM-based chatbot's personality evaluation. Our findings indicate that the chatbot's answers on human personality scales exhibit weak correlations with both user perception and interaction quality, which raises both criterion and predictive validity concerns of such a method. Further analysis revealed the role of task context and interaction in the chatbot's personality design assessment. We discuss the design implications for building contextualized and interactive evaluation of the chatbot's personality design
Optimization of Multiscale Compartmentalization Methods for Biotechnologies
No full textThe development of many biotechnologies requires the compartmentalization of biological components such as cells, protein, or nucleic acids into specialized compartments. Multiscale compartmentalization strategies are needed to handle multiple application scales including nanoscale drug delivery systems, microscale compartments for single-cell analysis, or milliscale compartments for tissue engineering. This work aims to develop methods for encapsulating multiple types of cargo into compartments ranging from the nanoscale to milliscale. The first section of this dissertation focuses on developing microfluidic droplet generation methods for high throughput and controllable encapsulation of cells into picoliter hydrogel particles. Challenges related to the integration of hydrogel with microfluidics, including on-chip crosslinking and hydrogel microparticle preservation, are addressed by optimizing input concentrations, flow rates, and collection methods. The preservation and functionality of the hydrogel materials in microscale form is validated to ensure compatibility with downstream applications, such as single cell analysis. Next, a milli-fluidic droplet generator is developed to create nanoliter sized droplets containing multiple cells. A scaled-up droplet generator was fabricated from a 3D printed mold using rapid prototyping principles to quickly develop a mold geometry to achieve multi-cell encapsulations in hydrogel. In the final section of the dissertation, nanoscale encapsulation of nucleic acids into membrane bound nanoparticles is investigated. Electroporation-mediated loading is used to introduce miRNA into extracellular vesicles, which are cell-derived membrane bound nanoparticles with high biocompatibility and therapeutic delivery potential. Immunocapture microbeads are utilized to select a subpopulation of the extracellular vesicles, while also improving the processing efficiency and purity, demonstrating the benefits of extracellular vesicles as nanoscale compartments for protecting and delivering exogenous RNA. Together, these works advance the understanding and applications of compartmentalization techniques, paving the way for more efficient biological assays across multiple scales
The Dynamics of Collective Action And Political Settlements: A Comparative Study of Somaliland and Somalia
No full textThis study examines the dynamics of collective action and political settlements in Somaliland and Somalia, emphasizing their divergent trajectories in state-building, peace-building, and governance during the critical period from 1991 to 2001. Despite their shared historical and cultural heritage, Somaliland and Somalia took vastly different paths following the collapse of Somalia's central government in 1991. Employing a process-tracing comparative framework, the research investigates how elite coalitions, traditional governance structures, external actors, and grassroots initiatives shaped political settlements in these contexts.
Somaliland, which had voluntarily united with Somalia in 1960 after gaining independence from the United Kingdom, reclaimed its sovereignty in 1991 in response to decades of marginalization and conflict. While southern Somalia descended into prolonged chaos, Somaliland achieved notable stability through the integration of traditional clan-based systems with modern governance mechanisms. Its localized, inclusive approach to peace-building fostered internal cohesion and sustainability. Conversely, Somalia’s political settlements, heavily influenced by external actors, often excluded local stakeholders, undermining collective action and sustainable governance.
The research also explores the roles of regional and international actors in shaping political outcomes, analyzing how Somaliland capitalized on its autonomy to establish effective governance, while Somalia’s reliance on externally driven models exacerbated internal divisions. Drawing on theories of collective action, political settlements, the politics of rents, and development gambling, this study unpacks the interplay of governance structures, elite power dynamics, economic incentives, and security challenges in these contexts.
The findings highlight Somaliland’s success in building cohesive local governance structures, which contributed to its relative stability, and contrast this with Somalia’s fragmented approach, which hindered progress. By comparing these cases, the research offers critical insights into the factors that enable or impede peace-building and state-building in fragile settings. It underscores the importance of fostering inclusive political settlements that prioritize local agency, respect indigenous governance traditions, and align with the socio-political dynamics of post-conflict societies. Ultimately, the study contributes to broader debates on sustainable pathways for governance and development in Africa and other conflict-affected regions
Hierarchical Computations of Social Interaction Perception in the Human Mind and Brain
No full textSeeing the interactions between other people is a critical part of our everyday visual experience, but recognizing the social interactions of others is often considered outside the scope of vision and grouped with higher level social cognition like theory of mind. I first review behavioral, computational, and neuroimaging evidence that social interaction perception is efficient and automatic, occurs in visually--selective regions of the brain, and is well modeled by bottom--up computational algorithms. We propose a computational framework and then present empirical evidence for this theory. Using fMRI, we find that features of social interactions are represented hierarchically along a posterior--to--anterior axis in lateral visual regions of the brain. These features include low--level visual features, mid-level social primitives (or visual spatial relations among people), and high-level social interactions among people. We then evaluate whether state--of--the--art (SOTA) computational algorithms have the right kinds of computations to model social interaction perception. By evaluating over 350 image, video, and language models, we find that dynamic and relational representations are necessary to model social interaction perception. However, we also identify a significant gap between the performance of SOTA algorithms and human perception and a lack of hierarchical correspondence between models and the brain. Finally, we use EEG--fMRI fusion to investigate the spatiotemporal dynamics of social interaction perception in the human brain. As in fMRI, we find that there is a temporal hierarchy in which low--level visual features are represented before mid-- and high--level social features. However, we find that the latency in mid-- and high--level lateral visual regions are comparable suggesting that mid-level representations may not be required for high-level representations of social interactions. Together, these studies provide strong evidence that social action features, particularly communicative interactions, are extracted along the lateral stream, though also suggest some key differences compared to the hierarchy in the ventral visual stream