137091 research outputs found
Sort by
Efficient Image Compression for Embedded Visual Odometry
No full textIn this paper, we explore the application of singular value based decomposition for image compression in the context of visual odometry.UndergraduateComputer Scienc
Non-invasive Arc Duration Measurement Based on Different Physical Emissions
No full textThe objective of this research is to develop a robust online non-invasive arc duration measurement method in the substation, which is a critical factor for circuit breaker online contact erosion estimation and implementation of condition-based and reliability-centered maintenance practices. In this research, different arc duration measurement methods using various physical signals, including very-low-frequency and low-frequency magnetic field, vibration, and sound signals are proposed. Each method, however, has its limitations. For the method based on very-low-frequency and low-frequency magnetic field, although the waveform signatures can indicate the arc initiation and extinction time, their susceptibility to contact geometry changes caused by contact erosion restricts the application scenarios of this method. The domain-adaptive method based on vibration and sound signals, on the other hand, can achieve a high accuracy (below 0.2 ms) comparable to existing techniques and exhibit robustness against noise interference and domain shift, which existing methods lack. However, the method requires representative substation data to train neural networks. Eventually, a decision-level fusion strategy is proposed. This strategy can combine existing and proposed methods to address the limitations that cannot be solved by a single method individually and enhance the accuracy and robustness of arc duration measurement in the substation.Ph.D.Electrical and Computer Engineerin
3D Fibrin Microgel System for Senescense Detection
No full textWith the scientific advances and improvements in quality of life over the past century, the percentage of the aged population has been steadily increasing. With age comes a multitude of age-related diseases including cardiovascular disease, hypertension, cancer, osteoporosis, Alzheimer’s disease, and more. One of the 12 hallmarks of aging, cellular senescence, has been directly linked to age-related pathologies. Treatment of cellular senescence via senolytics is a growing niche in aging research, due to the promise it shows in managing senescence accumulation and age-related diseases. Despite growing use of 3D systems in research, there is a lack of models for age-related drug screening. This thesis aims to refine a bio-printed fibroblast-containing fibrin gel model for the detection of cellular senescence. This fibrin microgel model closely mimics the early stages of wound healing, where clot formation occurs, and is subsequently degraded by a process known as fibrinolysis. First, we sought to establish reproducibility in the system. Gel parameters were studied to determine their relationship with an established readout of fibrinolysis time, and ways to reduce variability in the system were investigated. Second, different methods of senescence were tested to determine sensitivity of the fibrin gel model to cellular senescence. Comparing the effects of multiple forms of senescence in the model is necessary due to different pathways of senescence activation and potential differences in senescence associated secretory phenotypes (SASPs). Overall, the work accomplished in this thesis could help establish a novel marker for rapid cellular senescence detection and a tool for senolytic drug discovery and screening.M.S.Bioengineerin
Leveraging Prior Pedagogy: A Case Study of an Out-of-Field Teacher Adapting to After-School CS
No full textAfter-school programs are vital for broadening STEM participation, particularly for underrepresented youth, yet face significant staffing shortages, especially in Computer Science (CS). This thesis investigates the potential of hiring out-of-field (OOF) teachers to address this gap through a qualitative case study of one elementary school teacher with no prior CS experience teaching in an after-school CS program.
Using classroom observations, instructor reflections, and semi-structured interviews, the study explored how the instructor's OOF status and prior pedagogical knowledge (PK) influenced her teaching within the informal after-school environment. Findings indicate that while the instructor faced challenges typical of OOF teaching, such as limited CS content knowledge, difficulty with technical vocabulary, and debugging, she successfully leveraged her extensive PK from teaching kindergarten. Strategies like strong classroom management, differentiation, adaptability, and connecting content to student interests proved crucial.
The unique after-school environment, characterized by student choice, prominent social dynamics, and the instructor holding multiple roles (teacher, coach), significantly shaped her approach, demanding a focus on engagement and relationship-building for program sustainability. The instructor's high teaching self-efficacy and strong PK appeared to mitigate the negative effects often associated with OOF teaching.
This study suggests that pedagogical expertise can be as critical as content knowledge for OOF instructors in after-school CS programs. It highlights the importance of student engagement for sustainability in voluntary learning settings and offers insights into how OOF teachers adapt existing knowledge, informing recruitment practices and professional development design for such programs.UndergraduateHuman – Computer Interactio
CHAT Junior - Dolphin Whistle Recognition on the Edge
No full textThe Cetacean Hearing Augmentation and Telemetry (CHAT) and Unsupervised Harvesting and Utilization of Recognizable Acoustics (UHURA) project works to study communication between humans and wild dolphins. The CHAT/UHURA researchers communicate with each other, and dolphins by playing and recognizing audio signals that are similar to whistles dolphins might emit. A platform for this function called Wear-A-CUDA (referred to as CHAT Senior) currently exists, but it suffers from higher power consumption, mobility restrictions, and a poor user interface. This paper presents a solution for these issues by creating an audio classification model that can be deployed on a mobile application that does not face the same issues. This solution, CHAT Junior, uses a convolutional neural network trained to recognize whistle audio signals for various objects that researchers and dolphins interact with underwater. This model takes advantage of powerful mobile devices that contain tensor processing units to compute model predictions quickly. This paper experiments with four model architectures of various sizes that preprocess incoming audio signals into spectrograms using a short-time Fourier transform and use a convolutional neural network to classify the spectrogram input. After evaluating the different architectures on both their accuracy and prediction speed, the largest model was chosen. The largest model had 80 percent accuracy on a continuous stream of audio data while executing each prediction in 40 milliseconds, half the initial target time of 80 milliseconds. Though it still has limitations in extremely noisy environments or when whistle signals look very similar to each other, it performs well in the field and consistently differentiates between whistle signal classes and background noise. The model is now deployed in an Android mobile application, which uses an intuitive user interface to provide a better experience for CHAT/UHURA researchers in the field.UndergraduateComputer Scienc
Integrative computational-experimental study of the biomechanics of Cerebral Cavernous Malformation and its pharmacological rescue
No full textCerebral Cavernous Malformations (CCMs) are vascular abnormalities in the brain and spinal cord characterized by clusters of dilated capillaries with thin, fragile walls prone to leaking. These vascular lesions can lead to critical neurological issues such as seizures, focal neurological deficits, and potentially fatal brain hemorrhages. Current treatment options are predominantly surgical, involving the removal of the accessible regions of the affected tissue. However, surgery carries significant risks, especially for deep-seated lesions in critical brain areas. Moreover, not all patients are suitable candidates for surgery, and those who undergo surgery can develop new lesions over time. Despite advances in medical research, there remains a pressing need for non-surgical treatment options that can effectively manage or cure CCM disease.
Identifying pharmaceutical strategies is crucial for providing safer, more accessible treatments that can address the underlying molecular and cellular causes of CCMs, prevent lesion progression, and improve the quality of life for affected individuals. This thesis aims to unveil the dysregulations of molecular pathways and biomechanical deficiencies in cell function, resulting from CCM mutations and suggests two small-molecular inhibitors as novel pharmacological agents for the disease intervention.
To this end, we developed a multi-scale computational model to reproduce healthy, diseased, pharmacologically rescued endothelial cell behaviors during vascular network formation, accounting for the viscoelastic properties of cell bodies, mechanosensing, dynamics of cellular protrusions, and response of cell contacts to mechanical loading. The model replicates distinct multicellular patterns observed in wild-type and CCM knockdown cultures, revealing the critical role of the balance between cell-cell and cell-extracellular matrix (ECM) interactions in maintaining vascular integrity. Experimental validation using the tube formation assay confirmed the model's predictions, showing that CCM1 knockdown disturbs the balance by affecting mainly cell-ECM adhesion, while CCM3 knockdown disrupts the balance by significantly weakening cell-cell contacts.
Our RNA-seq analysis of CCM phenotypes identified significant changes in gene expression profiles, particularly in genes related to cell adhesion, ECM organization, and cytoskeletal regulation. The key targets identified by the analysis include genes associated with the WNT signaling pathway, such as AKT2 and GNAI2. To validate these targets, we cross-referenced our findings with publicly available datasets from CCM patient samples and CCM3-deficient mouse models. This comparative analysis confirmed the relevance of the WNT pathway and the involvement of AKT2 and GNAI2 in CCM pathology.
Finally, we employed pharmacological inhibitors to rescue the severely disrupted patterns in CCM1 and CCM3 knockdown cultures. LY20909314, a GSK3B inhibitor similar in its action to AKT2, and XAV939, a Tankyrase inhibitor that stabilizes Axin similar to GNAI2, were used to rescue both CCM phenotypes. These inhibitors demonstrated a high efficiency in restoring normal (wild-type) collective behavior of the CCM1 and CCM3 knockdown cells.
In conclusion, this thesis provides a comprehensive mechanobiological framework for understanding the molecular and cellular mechanisms driving CCM pathology. By integrating computational modeling, RNA-seq analysis, and experimental validation, we identified novel therapeutic candidates and demonstrated their efficiency in cell cultures. Our findings suggest that targeting CCM rescue through WNT signaling modulation may offer effective strategies for treating CCM, ultimately improving patient outcomes and reducing reliance on surgical interventions.Ph.D.Bioinformatic
Smart Charging of Electric Vehicles: Consumer Costs and Demand Reduction with Solar Power and Energy Storage
No full textGlobal efforts to reduce harmful emissions by internal combustion engines have served as motivation to electrify the transportation sector. In the United States (US), a variety of state-specific incentive programs and the increasing affordability of battery technologies have promoted the penetration of EVs in the market. Still, an increasing presence of EVs in grid networks brings the potential for significant stresses on power distribution infrastructure (voltage violations and/or transformer overloads), especially as overnight charging patterns persist. Simultaneously, the US is seeing a rise in solar photovoltaic (PV) power capacity, generating similar concerns over power/voltage quality. As a result, the rise of EVs necessitates charging strategies that minimize operation costs, maximize carbon-free energy (CFE) consumption, and minimize peak demands, while the rise of CFE generation behind electric meters has encouraged investments into battery energy storage systems (BESS) to more optimally utilize solar power. Smart charging (SC) provides an inexpensive approach to managing EV charging loads and CFE utilization in the near-term, addressing the challenges associated with rising power demands. Novel implementation of SC methods with consideration of PV and BESS behind the meter could support the promising emissions reductions offered by EVs while satisfying the needs of both consumers (e.g., EV owners, fleet owners, aggregators) and utility operators. In the proposed thesis, smart charging will be explored to quantify cost and grid impacts for i) residential homes with a single EV, BESS, and PV, ii) EV fleets that utilize CFE such as PV and BESS, and iii) an in-depth case study analyzing the electrification of public school buses and the potential to charge smart.M.S.Mechanical Engineerin
Machine Learning for Modeling Progression and Heterogeneity in Alzheimer's Disease
No full textMachine Learning for Modeling Progression and Heterogeneity in Alzheimer’s Disease
Raghav Tandon
185 Pages
Directed by Dr. Cassie S. Mitchell
This thesis uses existing and novel machine learning algorithms to study three related questions in Alzheimer’s disease (AD) research. These questions are - 1) What are the proteomic markers of change in AD? 2) In what sequence do changes take place as AD progresses?, and 3) What are the heterogeneities in disease progression among the AD population?
These questions are important due to some key challenges faced in AD research. Disease aetiology in AD is incompletely understood which has prevented the development of effective therapies. Though recent drug developments have resulted in disease modifying therapies (Lecanemab), these drugs have been found to be effective in early disease stages. Statistics show that early AD detection is a challenge and only a small fraction of people with AD are diagnosed early enough when available disease modifying treatments can show effectiveness. Further, AD is known to have a heterogeneous clinical presentation which complicates its clinical management. Finally, given the long disease evolution timelines which span several decades, collecting data throughout the disease evolution process has practical challenges and limitations.
The three questions this thesis aims to address are relevant to the above challenges. Identifying proteomic markers of change in the disease helps to elucidate the proteopathic changes underlying AD. The identified markers also have diagnostic utility and can be potentially useful for early disease detection. Understanding the sequence of progressive changes helps to develop new disease staging systems which are data driven and serve as early warning systems for the disease. Studying heterogeneities in AD progression helps to understand variations in important clinical variables such as rate of progression and cognitive decline, comorbidities, age of onset, and markers of disease pathology. This understanding about disease heterogeneity can be useful in two important ways. First, it can aid clinical decision making which helps to move towards a personalized medicine approach in AD. Second, it can be useful in clinical trial designs to control for disease related phenotypic heterogeneities and making the trial more sensitive to treatment effects.
These questions are answered using machine learning models applied to cross-sectional data, which makes the approaches data economical. Despite using limited data, the learned models generalize well to new data from external sources and future follow up visits from patients, which shows the potential utility of these approaches in solving important clinical challenges. The first question (“markers of change?”) is addressed using classic supervised learning approaches. The second (“sequence of disease related changes?”) develops new probabilistic generative algorithms which attempt to construct a disease progression trajectory from a large number of disease related neuroimaging biomarkers. Finally, the third question (“heterogeneities in progression”) extends the previous generative algorithm to model disease progression to take place over one of potentially many disease trajectories.
This thesis makes two main contributions. First, it contributes to a better understanding of AD. More specifically, these are –
1. Identifying the role of dysfunctional sugar metabolism in disease development.
2. Identifying important peptide markers which can be useful for early disease detection.
3. Inferring a trajectory of disease related neuroanatomical changes which can be useful in disease staging.
4. Understanding the heterogeneities in disease progression rates, brain regions affected and cognitive performance which can aid clinical decision making.
The second contribution of this thesis is to introduce new ML algorithms to model
disease progression in AD. These algorithms are –
1. scaled Event Based Model (sEBM): A probabilistic generative approach which models disease progression as a sequence of biomarker abnormalities. It extends previous algorithms by solving important computational challenges and makes the resulting algorithm scalable to a much larger number of disease markers.
2. scaled Subtype and Stage Inference (s-SuStaIn): Extends sEBM to infer varying disease progression trajectories. Experiments show that including a larger set of disease features to model progression can be helpful in identifying significantly varying disease progression trajectories.
While these algorithms have been applied to AD in this thesis, these are equally applicable to other neurodegenerative conditions such as Parkinson’s Disease, frontotemporal dementia, and progressive supranuclear palsy.Ph.D.Machine Learnin
Fourier Light-Field Microscopy: Design, Optimization, and Applications
No full textVisualizing diverse anatomical and functional traits that span many spatial scales with high spatiotemporal resolution provides insights into the fundamentals of biological systems. Light-field
microscopy (LFM) has recently emerged as a scanning-free, scalable method allowing for highspeed, volumetric imaging ranging from single-cell specimens to the mammalian brain. However,
LFM is far prevented from broader applications by the prohibitive reconstruction artifacts and
severe computational cost due to its nonuniform axial sampling and laterally variant PSF.
To address the challenge, in this thesis, we report Fourier LFM (FLFM), a system that processes
the light-field information through the Fourier domain, substantially overcomes the drawbacks of
LFM and realizes fast 3D imaging with enhanced spatial resolution and extended imaging depth.
Starting from the basic principle of optics, we established a complete framework consisting of
theory and algorithm for the light propagation, image formation, volume reconstruction, and
system characterization of FLFM. We proposed the generic rules for system design and
optimization, following which we developed two FLFM systems optimized for fast, 3D, live
imaging on, respectively, subcellular activities of organelles and functional and morphological
change of whole organoids. Based on these works as a solid base, we first, in Aim1, introduce
simultaneous two-color Fourier light-field imaging. By equipping a filter array behind the MLA
and allocating elemental images of imaged samples into different channels, we redistribute
fluorescent emission among spatial, angular, and chromatic dimensions and enable simultaneously
sensing ultrafast biological signals from two channels, breaking the camera limit on dual-color
FLFM imaging with alternative illumination. Then, in Aim 2, we developed an algorithm to
accelerate the reconstruction speed of FLFM by reducing the iterations in 3D deconvolution with
projection estimation, compressing 80%~90% of the time cost of reconstructing a volume. In the
end, in Aim3, we employed this IDC-FLFM system equipped with an accelerated reconstruction
algorithm in diverse biomedical studies, such as interrogating the mechanism of cancer invasion
in mammary organoids, toxicity assay on drugs for cardiac treatment, gene screening on
cardiovascular development, and detecting neural connections in animal locomotion. The imaging
results demonstrated the imaging capacity of our system for high-throughput screening and fast
time-lapsed 3D recording of rapid morphological change and functional traits, promising FLFM a
potent tool for imaging diverse phenotypic and functional information spanning broad molecular,
cellular, and tissue systems.Ph.D.Biomedical Engineerin