Matching Items (141)
Filtering by
Description
Video object segmentation (VOS) is an important task in computer vision with a lot of applications, e.g., video editing, object tracking, and object based encoding. Different from image object segmentation, video object segmentation must consider both spatial and temporal coherence for the object. Despite extensive previous work, the problem is still challenging. Usually, foreground object in the video draws more attention from humans, i.e. it is salient. In this thesis we tackle the problem from the aspect of saliency, where saliency means a certain subset of visual information selected by a visual system (human or machine). We present a novel unsupervised method for video object segmentation that considers both low level vision cues and high level motion cues. In our model, video object segmentation can be formulated as a unified energy minimization problem and solved in polynomial time by employing the min-cut algorithm. Specifically, our energy function comprises the unary term and pair-wise interaction energy term respectively, where unary term measures region saliency and interaction term smooths the mutual effects between object saliency and motion saliency. Object saliency is computed in spatial domain from each discrete frame using multi-scale context features, e.g., color histogram, gradient, and graph based manifold ranking. Meanwhile, motion saliency is calculated in temporal domain by extracting phase information of the video. In the experimental section of this thesis, our proposed method has been evaluated on several benchmark datasets. In MSRA 1000 dataset the result demonstrates that our spatial object saliency detection is superior to the state-of-art methods. Moreover, our temporal motion saliency detector can achieve better performance than existing motion detection approaches in UCF sports action analysis dataset and Weizmann dataset respectively. Finally, we show the attractive empirical result and quantitative evaluation of our approach on two benchmark video object segmentation datasets.
ContributorsWang, Yilin (Author) / Li, Baoxin (Thesis advisor) / Wang, Yalin (Committee member) / Cleveau, David (Committee member) / Arizona State University (Publisher)
Created2013
Description
Learning from high dimensional biomedical data attracts lots of attention recently. High dimensional biomedical data often suffer from the curse of dimensionality and have imbalanced class distributions. Both of these features of biomedical data, high dimensionality and imbalanced class distributions, are challenging for traditional machine learning methods and may affect the model performance. In this thesis, I focus on developing learning methods for the high-dimensional imbalanced biomedical data. In the first part, a sparse canonical correlation analysis (CCA) method is presented. The penalty terms is used to control the sparsity of the projection matrices of CCA. The sparse CCA method is then applied to find patterns among biomedical data sets and labels, or to find patterns among different data sources. In the second part, I discuss several learning problems for imbalanced biomedical data. Note that traditional learning systems are often biased when the biomedical data are imbalanced. Therefore, traditional evaluations such as accuracy may be inappropriate for such cases. I then discuss several alternative evaluation criteria to evaluate the learning performance. For imbalanced binary classification problems, I use the undersampling based classifiers ensemble (UEM) strategy to obtain accurate models for both classes of samples. A small sphere and large margin (SSLM) approach is also presented to detect rare abnormal samples from a large number of subjects. In addition, I apply multiple feature selection and clustering methods to deal with high-dimensional data and data with highly correlated features. Experiments on high-dimensional imbalanced biomedical data are presented which illustrate the effectiveness and efficiency of my methods.
ContributorsYang, Tao (Author) / Ye, Jieping (Thesis advisor) / Wang, Yalin (Committee member) / Davulcu, Hasan (Committee member) / Arizona State University (Publisher)
Created2013
Description
Microfluidic platforms have been exploited extensively as a tool for the separation of particles by electric field manipulation. Microfluidic devices can facilitate the manipulation of particles by dielectrophoresis. Separation of particles by size and type has been demonstrated by insulator-based dielectrophoresis in a microfluidic device. Thus, manipulating particles by size has been widely studied throughout the years. It has been shown that size-heterogeneity in organelles has been linked to multiple diseases from abnormal organelle size. Here, a mixture of two sizes of polystyrene beads (0.28 and 0.87 μm) was separated by a ratchet migration mechanism under a continuous flow (20 nL/min). Furthermore, to achieve high-throughput separation, different ratchet devices were designed to achieve high-volume separation. Recently, enormous efforts have been made to manipulate small size DNA and proteins. Here, a microfluidic device comprising of multiple valves acting as insulating constrictions when a potential is applied is presented. The tunability of the electric field gradient is evaluated by a COMSOL model, indicating that high electric field gradients can be reached by deflecting the valve at a certain distance. Experimentally, the tunability of the dynamic constriction was demonstrated by conducting a pressure study to estimate the gap distance between the valve and the substrate at different applied pressures. Finally, as a proof of principle, 0.87 μm polystyrene beads were manipulated by dielectrophoresis. These microfluidic platforms will aid in the understanding of size-heterogeneity of organelles for biomolecular assessment and achieve separation of nanometer-size DNA and proteins by dielectrophoresis.
ContributorsOrtiz, Ricardo (Author) / Ros, Alexandra (Thesis advisor) / Hayes, Mark (Committee member) / Borges, Chad (Committee member) / Arizona State University (Publisher)
Created2021
Description
This dissertation constructs a new computational processing framework to robustly and precisely quantify retinotopic maps based on their angle distortion properties. More generally, this framework solves the problem of how to robustly and precisely quantify (angle) distortions of noisy or incomplete (boundary enclosed) 2-dimensional surface to surface mappings. This framework builds upon the Beltrami Coefficient (BC) description of quasiconformal mappings that directly quantifies local mapping (circles to ellipses) distortions between diffeomorphisms of boundary enclosed plane domains homeomorphic to the unit disk. A new map called the Beltrami Coefficient Map (BCM) was constructed to describe distortions in retinotopic maps. The BCM can be used to fully reconstruct the original target surface (retinal visual field) of retinotopic maps. This dissertation also compared retinotopic maps in the visual processing cascade, which is a series of connected retinotopic maps responsible for visual data processing of physical images captured by the eyes. By comparing the BCM results from a large Human Connectome project (HCP) retinotopic dataset (N=181), a new computational quasiconformal mapping description of the transformed retinal image as it passes through the cascade is proposed, which is not present in any current literature. The description applied on HCP data provided direct visible and quantifiable geometric properties of the cascade in a way that has not been observed before. Because retinotopic maps are generated from in vivo noisy functional magnetic resonance imaging (fMRI), quantifying them comes with a certain degree of uncertainty. To quantify the uncertainties in the quantification results, it is necessary to generate statistical models of retinotopic maps from their BCMs and raw fMRI signals. Considering that estimating retinotopic maps from real noisy fMRI time series data using the population receptive field (pRF) model is a time consuming process, a convolutional neural network (CNN) was constructed and trained to predict pRF model parameters from real noisy fMRI data
ContributorsTa, Duyan Nguyen (Author) / Wang, Yalin (Thesis advisor) / Lu, Zhong-Lin (Committee member) / Hansford, Dianne (Committee member) / Liu, Huan (Committee member) / Li, Baoxin (Committee member) / Arizona State University (Publisher)
Created2022
Description
Traditional Reinforcement Learning (RL) assumes to learn policies with respect to reward available from the environment but sometimes learning in a complex domain requires wisdom which comes from a wide range of experience. In behavior based robotics, it is observed that a complex behavior can be described by a combination of simpler behaviors. It is tempting to apply similar idea such that simpler behaviors can be combined in a meaningful way to tailor the complex combination. Such an approach would enable faster learning and modular design of behaviors. Complex behaviors can be combined with other behaviors to create even more advanced behaviors resulting in a rich set of possibilities. Similar to RL, combined behavior can keep evolving by interacting with the environment. The requirement of this method is to specify a reasonable set of simple behaviors. In this research, I present an algorithm that aims at combining behavior such that the resulting behavior has characteristics of each individual behavior. This approach has been inspired by behavior based robotics, such as the subsumption architecture and motor schema-based design. The combination algorithm outputs n weights to combine behaviors linearly. The weights are state dependent and change dynamically at every step in an episode. This idea is tested on discrete and continuous environments like OpenAI’s “Lunar Lander” and “Biped Walker”. Results are compared with related domains like Multi-objective RL, Hierarchical RL, Transfer learning, and basic RL. It is observed that the combination of behaviors is a novel way of learning which helps the agent achieve required characteristics. A combination is learned for a given state and so the agent is able to learn faster in an efficient manner compared to other similar approaches. Agent beautifully demonstrates characteristics of multiple behaviors which helps the agent to learn and adapt to the environment. Future directions are also suggested as possible extensions to this research.
ContributorsVora, Kevin Jatin (Author) / Zhang, Yu (Thesis advisor) / Yang, Yezhou (Committee member) / Praharaj, Sarbeswar (Committee member) / Arizona State University (Publisher)
Created2021
Description
Retinotopic map, the map between visual inputs on the retina and neuronal activation in brain visual areas, is one of the central topics in visual neuroscience. For human observers, the map is typically obtained by analyzing functional magnetic resonance imaging (fMRI) signals of cortical responses to slowly moving visual stimuli on the retina. Biological evidences show the retinotopic mapping is topology-preserving/topological (i.e. keep the neighboring relationship after human brain process) within each visual region.
Unfortunately, due to limited spatial resolution and the signal-noise ratio of fMRI, state of art retinotopic map is not topological.
The topic was to model the topology-preserving condition mathematically, fix non-topological retinotopic map with numerical methods, and improve the quality of retinotopic maps.
The impose of topological condition, benefits several applications.
With the topological retinotopic maps, one may have a better insight on human retinotopic maps, including better cortical magnification factor quantification, more precise description of retinotopic maps, and potentially better exam ways of in Ophthalmology clinic.
ContributorsTu, Yanshuai (Author) / Wang, Yalin (Thesis advisor) / Lu, Zhong-Lin (Committee member) / Crook, Sharon (Committee member) / Yang, Yezhou (Committee member) / Zhang, Yu (Committee member) / Arizona State University (Publisher)
Created2022
Description
In cold chain tracking systems, accuracy and flexibility across different temperatures ranges plays an integral role in monitoring biospecimen integrity. However, while two common cold chain tracking systems are currently available (electronic and physics/chemical), there is not an affordable cold chain tracking mechanism that can be applied to a variety of temperatures while maintaining accuracy for individual vials. Hence, our lab implemented our understanding of biochemical reaction kinetics to develop a new cold chain tracking mechanism using the permanganate/oxalic acid reaction. The permanganate/oxalic acid reaction is characterized by the reduction of permanganate (MnVII) to Mn(II) with Mn(II)-autocatalyzed oxidation of oxalate to CO2, resulting in a pink to colorless visual indicator change when the reaction system is not in the solid state (i.e., frozen or vitrified). Throughout our research, we demonstrate, (i) Improved reaction consistency and accuracy along with extended run times with the implementation of a nitric acid-based labware washing protocol, (ii) Simulated reaction kinetics for the maximum length reaction and 60-minute reaction based on previously developed MATLAB scripts (iii) Experimental reaction kinetics to verify the simulated MATLAB maximum and 60-minute reactions times (iv) Long-term stability of the permanganate/oxalic acid reaction with water or eutectic solutions of sodium perchlorate and magnesium perchlorate at -80°C (v) Reaction kinetics with eutectic solvents, sodium perchlorate and magnesium perchlorate, at 25°C, 4°C, and -8°C (vi) Accelerated reaction kinetics after the addition of varying concentrations of manganese perchlorate (vii) Reaction kinetics of higher concentration reaction systems (5x and 10x; for darker colors), at 25°C (viii) Long-term stability of the 10x higher concentration reaction at -80°C.
ContributorsLjungberg, Emil (Author) / Borges, Chad (Thesis director) / Levitus, Marcia (Committee member) / Williams, Peter (Committee member) / Barrett, The Honors College (Contributor) / School of Molecular Sciences (Contributor) / Department of Psychology (Contributor)
Created2022-12
Description
Mycobacterial infections, as represented by leprosy and tuberculosis, have persisted as human pathogens for millennia. Their environmental counterparts, nontuberculous mycobacteria (NTM), are commodious infectious agents endowed with extensive innate and acquired antimicrobial resistance. The current drug development process selects for antibiotics with high specificity for definitive targets within bacterial metabolic and replication pathways. Because these compounds demonstrate limited efficacy against mycobacteria, novel antimycobacterial agents with unconventional mechanisms of action were identified. Two highly resistant NTMs, Mycobacterium abscessus (Mabs) a rapid-growing respiratory, skin, and soft tissue pathogen, and Mycobacterium ulcerans (MU), the causative agent of Buruli ulcer, were selected as targets. Compounds that indicated antimicrobial activity against other highly resistant pathogens were selected for initial screening. Antimicrobial peptides (AMPs) have demonstrated activity against a variety of bacterial pathogens, including mycobacterial species. Designed antimicrobial peptides (dAMPs), rationally-designed and synthetic contingents, combine iterative features of natural AMPs to achieve superior antimicrobial activity in resistant pathogens. Initial screening identified two dAMPs, RP554 and RP557, with bactericidal activity against Mabs. Clay-associated ions have previously demonstrated bactericidal activity against MU. Synthetic and customizable aluminosilicates have also demonstrated adsorption of bacterial cells and toxins. On this basis, two aluminosilicate materials, geopolymers (GP) and ion-exchange nanozeolites (IE-nZeos), were screened for antimicrobial activity against MU and its fast-growing relative, Mycobacterium marinum (Mmar). GPs demonstrated adsorption of MU cells and mycolactone, a secreted, lipophilic toxin, whereas Cu-nZeos and Ag-nZeos demonstrated antibacterial activity against MU and Mmar. Cumulatively, these results indicate that an integrative drug selection process may yield a new generation of antimycobacterial agents.
ContributorsDermody, Roslyn June (Author) / Haydel, Shelley E (Thesis advisor) / Bean, Heather (Committee member) / Nickerson, Cheryl (Committee member) / Stephanopoulos, Nicholas (Committee member) / Arizona State University (Publisher)
Created2022
Description
Plasma and serum are the most commonly used liquid biospecimens in biomarker research. These samples may be subjected to several pre-analytical variables (PAVs) during collection, processing and storage. Exposure to thawed conditions (temperatures above -30 °C) is a PAV that is hard to control, and track and could provide misleading information, that fail to accurately reveal the in vivo biological reality, when unaccounted for. Hence, assays that can empirically check the integrity of plasma and serum samples are crucial. As a solution to this issue, an assay titled ΔS-Cys-Albumin was developed and validated. The reference range of ΔS-Cys-Albumin in cardio vascular patients was determined and the change in ΔS-Cys-Albumin values in different samples over time course incubations at room temperature, 4 °C and -20 °C were evaluated. In blind challenges, this assay proved to be successful in identifying improperly stored samples individually and as groups. Then, the correlation between the instability of several clinically important proteins in plasma from healthy and cancer patients at room temperature, 4 °C and -20 °C was assessed. Results showed a linear inverse relationship between the percentage of proteins destabilized and ΔS-Cys-Albumin regardless of the specific time or temperature of exposure, proving ΔS-Cys-Albumin as an effective surrogate marker to track the stability of clinically relevant analytes in plasma. The stability of oxidized LDL in serum at different temperatures was assessed in serum samples and it stayed stable at all temperatures evaluated. The ΔS-Cys-Albumin requires the use of an LC-ESI-MS instrument which limits its availability to most clinical research laboratories. To overcome this hurdle, an absorbance-based assay that can be measured using a plate reader was developed as an alternative to the ΔS-Cys-Albumin assay. Assay development and analytical validation procedures are reported herein. After that, the range of absorbance in plasma and serum from control and cancer patients were determined and the change in absorbance over a time course incubation at room temperature, 4 °C and -20 °C was assessed. The results showed that the absorbance assay would act as a good alternative to the ΔS-Cys-Albumin assay.
ContributorsJehanathan, Nilojan (Author) / Borges, Chad (Thesis advisor) / Guo, Jia (Committee member) / Van Horn, Wade (Committee member) / Arizona State University (Publisher)
Created2022
Description
Neural tissue is a delicate system comprised of neurons and their synapses, glial cells for support, and vasculature for oxygen and nutrient delivery. This complexity ultimately gives rise to the human brain, a system researchers have become increasingly interested in replicating for artificial intelligence purposes. Some have even gone so far as to use neuronal cultures as computing hardware, but utilizing an environment closer to a living brain means having to grapple with the same issues faced by clinicians and researchers trying to treat brain disorders. Most outstanding among these are the problems that arise with invasive interfaces. Optical techniques that use fluorescent dyes and proteins have emerged as a solution for noninvasive imaging with single-cell resolution in vitro and in vivo, but feeding in information in the form of neuromodulation still requires implanted electrodes. The implantation process of these electrodes damages nearby neurons and their connections, causes hemorrhaging, and leads to scarring and gliosis that diminish efficacy. Here, a new approach for noninvasive neuromodulation with high spatial precision is described. It makes use of a combination of ultrasound, high frequency acoustic energy that can be focused to submillimeter regions at significant depths, and electric fields, an effective tool for neuromodulation that lacks spatial precision when used in a noninvasive manner. The hypothesis is that, when combined in a specific manner, these will lead to nonlinear effects at neuronal membranes that cause cells only in the region of overlap to be stimulated. Computational modeling confirmed this combination to be uniquely stimulating, contingent on certain physical effects of ultrasound on cell membranes. Subsequent in vitro experiments led to inconclusive results, however, leaving the door open for future experimentation with modified configurations and approaches. The specific combination explored here is also not the only untested technique that may achieve a similar goal.
ContributorsNester, Elliot (Author) / Wang, Yalin (Thesis advisor) / Muthuswamy, Jitendran (Committee member) / Towe, Bruce (Committee member) / Arizona State University (Publisher)
Created2022