22689 research outputs found
Sort by
NUCLEIC ACID BINDING BY SAMHD1 INDUCES LIQUID-LIQUID PHASE SEPARATION IN VITRO
Get PDFSAMHD1 (Sterile-α-motif and histidine-aspartic acid domain-containing protein 1) is a dNTPase enzyme exclusively found in vertebrates with the additional activity of binding to both ssRNA and ssDNA. SAMHD1 is the only enzyme in humans that directly converts all canonical dNTPs to the nucleoside and tripolyphosphate products. The enzyme has diverse cellular functions including restriction of viral infections, DNA repair, RNA homeostasis, DNA replication, telomere stability, and the cellular innate immune response to foreign and self-nucleic acids. Many of the cellular functions of SAMHD1 are known to be regulated by liquid-liquid phase separation (LLPS), or 'biomolecular condensates”, but there have been no studies of the molecular properties of SAMHD1 under conditions of molecular crowding that result in LLPS in the cell. In this study, we perform in vitro experiments establishing that SAMHD1 can form liquid-like condensates in the presence of the inert crowding agent PEG2000, and that condensate formation requires binding to ssDNA or ssRNA. A systematic study of the factors that impact SAMHD1 phase separation was investigated. These include the nucleic acid sequence, its length, and the affinity and dynamics of the SAMHD1 nucleic acid complex. Our findings indicate that in the absence of other cellular proteins that are known to interact with SAMHD1, the enzyme has the inherent property of phase separation in the presence of single-stranded nucleic acids. We discuss the implications of these in vitro findings in the regulation of the dNTPase and nucleic acid binding functions of SAMHD1 in cells
Foucault's Other Marx: Limits, Origins, and the Production of the Natural
Get PDFThe dissertation advocates a Foucauldian reading of Karl Marx’s critique of political economy. I build upon Michel Foucault’s neglected comments about Marx’s historical method and his reflections on the role of theory in social movements in order to argue that Marx’s critique of political economy should be read as a “topological and geological survey of the battlefield” – rather than an intervention into historical debates about how or why people should struggle against capitalism. Moreover, I advocate reading Capital as an immanent exposition of capitalism’s categories, relations, and subjectivities. Such a reading heeds Foucault’s injunction to study how power instantiates and reproduces itself rather than ask what power, a seemingly transhistorical category, is. As such, I read Capital as an inquiry into how capitalism immanently posits itself and make the case for a productive (rather than merely repressive) view of power in Marx’s mature work. One of the strengths of Marx’s mature work, I claim, is its attentiveness to capitalism’s reproduction of itself through the naturalization of concepts. I argue readers of Capital have not sufficiently grasped this point. By reading the critique of political economy through the distinction between the natural and the social, readers of Capital trap Marx in the bourgeois paradigm he sought to escape. I argue a Foucauldian reading of the text helps us escape this paradigm and encourages us to understand Marx as a theorist of the production of the distinction between the natural and the social
Robust 3D Recognition via Analysis-by-Synthesis
No full textImages are 2D observations of the 3D world we live in. Recognition of objects from those images is one of the most important tasks for computer vision. Currently, the majority of computer vision approaches recognize 2D images without modeling the imaging process from 3D space. These approaches learn from the distribution of the train 2D images and are normally sensitive to out-of-distribution (OOD) situations. Although we have developed an alternative approach that significantly improves the robustness of 2D vision approaches, we believe it is still limited. Motivated by cognitive studies of human vision, we propose an object recognition pipeline, which recognizes objects from 2D images by predicting the 3D configuration in the 3D space while modeling the imaging process. Our proposed approach learns approximate category-level object representation by a combination of 3D object geometry and discriminative neural features. One crucial contribution of our proposed approach is the mechanism that bridges the 2D image and 3D world, which is namely Render-and-Compare. Specifically, Render-and-Compare conducts Analysis-by-Synthesis via differentiating the rendering process that provides cues for retrieval of 3D configurations of the objects in an iterative optimization process. On the other hand, our approach builds a flexible object centric representation. Using variant object geometry representations, optimization strategies, rendering strategies, and neural representations, our approach can be applied to lots of vision recognition tasks, including object pose estimation, shape estimation, detection, amodal segmentation, part detection, classification, and few-shot pose learning. During my path to pursuing PhD, I, with collaboration of others, have developed this approach and extensively studied all components in the pipeline for boarding its application with better accuracy and robustness. The extensive experiments conducted on various datasets demonstrate our approach ability for real-world usage
Targeting Ferroptosis In Acute Respiratory Distress Syndrome Through Immunometabolism
No full textAcute respiratory syndrome (ARDS) is a highly inflammatory and life-threatening pulmonary injury. There are no pharmacologic therapies approved for treatment of ARDS, and treatment remains largely supportive with lung-protective ventilation. In a search for new therapeutic approaches, we explored changes in lung metabolism associated with ARDS. In lung samples from four distinct mouse models of indirect and direct lung injury ARDS, and in bronchiolar alveolar lavage fluid from patients with severe COVID-19, we found a consistent reduction of glutathione (GSH) abundance in compared to in healthy controls. This implied that alveolar epithelial cells are exposed to oxidative stress in ARDS, a view that was supported by our findings of increased total cellular reactive oxygen species (ROS), lipid peroxides and DNA damage (which can be caused by ROS) in these cells in a Streptococcus pneumoniae-infection model of ARDS. Increased lipid peroxides in this setting led us to hypothesize that alveolar epithelial cells were dying from ferroptosis. To examine this, we treated S. pneumoniae-infected mice with the ferroptosis inhibitor Liproxstatin -1. This resulted in a reduction in lipid peroxides in alveolar epithelial cells, the recovery of oxygen saturation in the blood and an increased chance of survival. A common treatment for ARDS is mechanical lung ventilation with oxygen. We speculated that this treatment may itself lead to oxidative stress and increased risk of ferroptosis, but did not find evidence to support this idea. Whether oxygen treatment increases alveolar epithelial cell ferroptosis in mice with ARDS is currently under study
View Enhancement of Arthroscopic Surgery Through Augmented Reality and Neural Radiance Fields
Get PDFThis thesis aims to explore a novel framework that combines Neural Radiance Field (NeRF) based reconstruction and Augmented Reality (AR) to overcome challenges and innovations in arthroscopy, a minimally invasive orthopedic procedure that is commonly used for the treatment of joints such as the shoulder, hip, and knee. One significant challenge in arthroscopic surgery is the requirement for surgeons to coordinate the surgical tools and the arthroscopic camera simultaneously. This coordination becomes particularly complex due to the limited space within which the arthroscopy is performed, severely restricting the maneuverability of the arthroscope and, consequently, the surgeon's field of view.
To address these issues, my work introduces "ARthroNeRF," an innovative framework that merges the capabilities of Neural Radiance Field (NeRF) based reconstruction with Augmented Reality (AR). This integration facilitates the creation of synthetic viewpoints directly from the perspective of surgical tools, providing additional information that can be used by the surgeon to identify relevant information.
To assess the effectiveness of ARthroNeRF, this work provides a user study involving 18 participants. During this study, participants were required to locate and touch hidden targets using the synthetic viewpoint provided by the surgical tool’s perspective. The findings from this study indicated that ARthroNeRF successfully offers precise additional visual information, potentially enhancing the training and performance of arthroscopic procedures. Moreover, the system effectively displays these reconstructed scenes through Microsoft HoloLens 2, incorporating AR to superimpose synthesized images alongside traditional arthroscopic video within the surgeon’s field of view. This capability presents a promising method to improve spatial awareness in scenarios where spatial constraints pose significant challenges
Vascularized Deinnvervated Muscle Target for motor unit based prosthetic limb control
No full textThis thesis focuses on addressing the significant challenges faced by individuals with traumatic limb loss in prosthetic limb control. Chapter 1 provides a comprehensive review of current neural prosthetic technologies, setting the foundation for subsequent research. Chapter 2 introduces the general methods employed throughout the thesis.
Chapter 3, the centerpiece of the thesis, investigates the natural bio-separator effect on vascularized denervated muscle targets (VDMT) to assess its potential for muscle reinnervation. This exploration aims to uncover insights into the physiological mechanisms underlying muscle reinnervation post-amputation, offering valuable guidance for prosthetic interventions. Building upon the findings from Chapter 3, Chapter 4 presents a novel approach termed Bi-VDMT, which involves surgically separating the tibial nerve and implanting them on different sites of the VDMT to enhance muscle reinnervation. Lastly, Chapter 5 addresses a crucial limitation in current wireless communication platforms for implantable reinnervated muscle systems—the transmission bandwidth—by investigating the impact of wavelet compression and electrode placement strategies on motor unit information preservation. Collectively, this multi-chapter investigation contributes to advancing the field of prosthetics by offering both theoretical insights and practical solutions aimed at improving the functionality and usability of prosthetic devices for individuals with limb loss
AN ANALYSIS ON THE FORMATION OF PILLOWED COMPONENTS DURING SMT PRODUCTION: THE EFFECTS OF AN ELEVATED SOLDER MASK THICKNESS
Get PDFAnnapolis Micro Systems (AMS) is a company centered around the production of printed circuit assemblies (PCA). Their current manufacturing process has seen an increase in the occurrences of a phenomenon called pillowing. Pillowing is a manufacturing defect in which a component on the PCA does not form a proper solder connection. The goal of this study is to determine the cause of this manufacturing defect. This was done by investigating each of the common causes, then identifying which one was responsible for the increase in occurrences at AMS. After process review and preliminary studies, only one common cause remained unaccounted for, the solder mask thickness. Through stored imaging techniques, the solder mask thickness of any previous productions could be analyzed. A comprehensive study on the effects of varying solder mask thicknesses revealed that an elevated solder mask thickness substantially contributed to the presence and severity of pillowing. Through these AOI measurements it was found that a solder mask thickness above 20 μm increased the odds of pillowing by more than 70%. Due to the accuracy level of AOI measurements an adjustment was necessary. This was achieved through the cross-sectioning of studied boards which provided an adjusted value of 31.22 μm. As a result, two main recommendations were substantiated. First, during production, each new lot should have the solder mask of the first board measured, with consideration for pillowing if the board measures above 20 μm. Second, new specifications should be sent to PCB manufacturers with target solder mask thicknesses being well below 31.22 μm and they should not accept any board that falls above 31.22 μm, making it the upper limit for PCB manufacturing. AMS should continue monitoring the solder mask thickness and pillowing relationship. By following the recommendations laid out in this study, AMS should see a vast decrease in the occurrence of this defect during their manufacturing process
The Impacts of Inorganic Arsenic Exposure on Human Macrophages
No full textAccording to the World Health Organization, 140 million people in over 70 countries have exposure to inorganic arsenic in their drinking water. Arsenic toxicity is a major health issue, and can lead to several serious diseases like cancer, diabetes, pulmonary diseases, cardiovascular disease, infant mortality, reduced immune function in children and reduced cognitive development. Arsenic also has impacts on the immune system. This thesis aims to develop a workflow to characterize and analyze the impact of arsenic on human immune cells through flow cytometry
IMPACT OF OUTCOME MEASUREMENTS AND CUTOFF SCORES ON IDENTIFYING PREDICTORS IN TREATMENT RESPONSE IN AN APATHY IN DEMENTIA TRIAL
Get PDFObjective: The Apathy in Dementia Methylphenidate Trial 2 (ADMET 2) showed the effect of methylphenidate in treating apathy in Alzheimer’s disease with heterogeneous response, using the Neuropsychiatric Inventory Apathy Subscale (NPI) measurement. I assessed the impact on identification for clinical response predictors of different outcome measurements and different cutoff scores for those measurements, including the Clinical Global Impression of Change (CGIC), the Dementia Apathy Interview and Rating (DAIR), and the Neuropsychiatric Inventory (NPI).
Design and Setting: The Apathy in Dementia Methylphenidate Trial 2 (ADMET 2) was a Phase III, randomized, placebo-controlled multi-center trial. Our work used univariate and multivariate analyses to pool from 23 clinical potential predictors of response to methylphenidate.
Participants: 99 patients assigned to the active treatment arm in the 200-patient ADMET 2 trial, conducted at 10 sites in the U.S. and Canada.
Methods: The primary outcome measurements included the Neuropsychiatric Inventory
Apathy Subscale (NPI apathy), the Clinical Global Impression of Change (CGIC) and the Dementia Apathy Interview and Rating (DAIR), each with 3 different cut-off scores. A two-step process that included a univariate and a multivariate logistic regression was used to determine predictors of response to methylphenidate.
Results: In the methylphenidate treatment group, 99 participants (33 females and 65 males) completed the 6-month follow-up. After developing prediction models utilizing three distinct
measures of apathy, the results showed very different groups of clinical predictors. Examination of specific cutoffs for each of the outcome measures showed that the models
were stable in NPI and DAIR models but not in CGIC model. The consistent predictors across all models were robust indicators, including age, current smoking status, and a history of major depression. The predictor of clinical response associated with current trazodone use was also a good indicator in the majority of models.
Conclusion: The selection of outcome measures and the determination of cutoff scores influence the identification of predictors. Different outcome measures and cutoff scores had a significant impact on the univariate prediction models, altering the selection of potential
predictors that qualified for inclusion in the multivariate models. Age, smoking status and major depression showed significant improvement among all NPI-A, CGIC and DAIR measures with methylphenidate. When assessing the ROC curve and AUC values, it became
evident that the DAIR model outperformed the NPI and CGIC models, as it exhibited a higher AUC, signifying superior performance
Emergent Language for Modeling of Hierarchical Brain Networks
Get PDFBrain networks are known to be hierarchically organized. Deep learning methods can be used to learn these hierarchical representations, but often suffer from being unexplainable “black boxes” or requiring large amounts of labeled data. To address these limitations, we propose Emergent Language Symbolic Autoencoder (ELSA), a weakly supervised deep learning framework that incorporates an Emergent Language (EL) approach. This method generates symbolic sentences that represent the data, thus enhancing the model's interpretability. We modify EL by incorporating our "progressive" loss functions which encourage the model to produce hierarchically consistent sentences. We use these sentences to create progressive reconstructions of brain networks that illustrate the model’s interpretation of the gradual refining of “broader” networks to “granular” networks. We also used these sentences to generate connectivity matrices between networks. Finally, we introduce an evaluation method to check for hierarchical consistency of the sentences. In summary, our end-to-end framework presents a comprehensive hierarchical model of brain networks, demonstrating a novel way to create, evaluate, and a few ways to extract insights