1,263 research outputs found
Robust Automatic Multi-Sperm Tracking in Time-Lapse Images
No full textHuman sperm cell counting, tracking and motility analysis is of significant interest to biologists studying sperm function and to medical practitioners evaluating male infertility. Today, the prevailing method for analyzing sperm at fertility clinics and research laboratories is laborious and subjective. Namely, the number and quality of sperm are often visually appraised by technicians using a microscope. Although total sperm count and sperm concentration can be reasonably estimated when standard protocols are applied, they have little diagnostic value except in identifying pathologically extreme abnormalities. More dynamic sperm swimming parameters such as curvilinear velocity (VCL), straight-line velocity (VSL), linearity of forward progression (LIN) and amplitude of lateral head displacement (ALH) are increasingly believed to have clinical significance in predicting infertility but are impossible for a human observer to visually discern. Expensive computer-assisted semen analysis (CASA) instruments are also sometimes used but are severely encumbered by crude ad-hoc tracking algorithms which cannot track sperm in close proximity or whose paths intersect and are typically limited to analyzing video clips of < 1 sec duration.In this thesis, we present a robust automatic multi-sperm tracking algorithm that can measure dynamic sperm motility parameters over time in pre-recorded time-lapse images. This effort is informed by progress in signal processing and target tracking technologies over the last three decades. Multi-target tracking algorithms originally developed for radar, sonar and video processing have addressed similar problems in other domains. In this thesis, we demonstrate that their methodologies can be used for sperm tracking and motility analysis. To resolve sperm measurement-to-track association conflicts, we applied and evaluated three multi-target tracking algorithms: the probabilistic data association filter (PDAF), the joint probabilistic data association filter (JPDAF) and the exact nearest neighbor extension to the JPDAF (ENN-JPDAF). We validated the accuracy of our tracking and motility analysis by using simulated sperm trajectories whose ground truth tracks were perfectly known. Using samples collected from five patients at a fertility clinic, we demonstrated automatic sperm detection and tracking even during challenging multi-sperm collision events. Combined analysis, testing and simulation support the use of probabilistic data association techniques robust automatic multi-sperm tracking. This method could provide fertility specialists with new data visualizations and interpretations previously impossible with existing laboratory protocols
Detection in distributed sensor networks
No full textThis thesis describes detection and communication algorithms for distributed sensor networks.In the first part of the thesis, we investigate a new architecture for distributed binary hypothesis detection by employing a Collision Resolution Algorithm (CRA), where all local sensors share a common channel to communicate with the decision fusion center. This architecture is important in the design of sensor fields, where a large number of distributed sensors share a single "emergency" channel.In the second part of the thesis, we discuss an industrial application of such a distributed detection system, namely, the LonWorks control network. We concentrate on the predictive p-persistent CSMA protocol implemented in the MAC layer of LonWorks protocol, which was proposed by the Echelon Corporation in the 1980s. In order to model this algorithm, we expand the CRA model developed in the first part to analyze variable-length messages. Predictions of the model are compared to an OPNET simulator of LonWorks, and to resultsfrom a physical network.Finally, we propose a direction-of-arrival (DOA) algorithm for sensor networks. It employs an improved polynomial rooting method using unitary transformations.Ph.D., Electrical Engineering -- Drexel University, 200
A mathematical model of glucose metabolism in hospitalized patients with diabetes and stress hyperglycemia
No full textThe human body employs several mechanisms to regulate the concentration of glucose in the bloodstream. The rates of glucose uptake and release from specific organs within the body are modulated directly by the concentrations of metabolites and hormones, and indirectly by the autonomic nervous system. The negative feedback relationship between glucose and the anabolic hormone, insulin, dominates the process of glycemic regulation. The binding of insulin to its receptor begins a cascade of intracellular events that increases glucose uptake into the liver and peripheral tissue and reduces glucose release from the liver. However, this mechanism can be overwhelmed during the acute stress typified by a moderate surgical procedure. Cellular damage and tissue trauma cause a surge in catabolic hormones and cytokines, leading to insulin resistance and marked hyperglycemia. The focus of this study is the mathematical descriptions of the processes that regulate glucose production and uptake. Such descriptions model the complex relationships between metabolites and hormones and their effects on glycemia. Descriptive models that are accurate and robust have the potential to guide the development of tools designed to manage glycemia in hospitalized patients with diabetes and stress-induced hyperglycemia. Specifically, we investigate the validity of a glucose metabolism model published by John Sorensen in a 1985 doctoral thesis. The model is a set of 22 first-order time-invariant nonlinear differential equations describing the interaction of glucose, insulin and glucagon and their effect on organ-level glucose uptake and release. We modified the model to incorporate recent experimental data, including data we have collected in clinical trials. The model was expanded to include a description of epinephrine and its effects on glycemia.Ph.D., Biomedical Engineering -- Drexel University, 200
Facsimile intrusion systems over IP networks
No full textIn this thesis, we investigate the security vulnerabilities present in sending a facsimile document over an IP network. We have developed and tested an intrusion mechanism, which is capable of intruding into the facsimile communication over the Public Switched Telephone Network (PSTN) section of the FOIP network without either communicating parties having any knowledge of intrusion taken place. Additionally, we describe the various techniques by which intrusion can take place over an IP network. Finally, we conclude by suggesting countermeasure to prevent the occurrence of such attacks on the FOIP network in the future.M.S., Electrical Engineering -- Drexel University, 200
High bit-rate digital communication through metal channels
No full textThe need to transmit digital information across metallic barriers arises frequently in industrial control applications. In some applications, the barrier can be penetrated with wiring, while in others this may not be possible. For example, metal bulkheads, pressure vessels, or pipelines may require a level of mechanical integrity that prohibits mechanical penetration. This study investigates the use of ultrasonic signaling for data transmission across metallic barriers, discusses the associated challenges, and analyzes several alternative communication system implementations.Several recent e orts have been made to develop through-metal ultrasonic communication systems, with approaches ranging widely in bitrate, complexity, and power requirements. The transceiver designs presented here are intended to cover a range of target applications. In systems having low data rate requirements, simple transceivers with low hardware/software complexity can be used. At high data rates, however, severe echoing in the ultrasonic channel leads to intersymbol interference. Reliable high speed communication therefore requires the use of channel equalizers, and results in a transceiver with higher hardware/software complexity.In this thesis, issues related to the design of reliable through-metal ultrasonic communication systems are discussed. These include (1) the development of mathematical models used to characterize the channel, (2) application of equalization techniques needed to achieve high-speed communication, and (3) analysis of hardware/software complexity for alternative transceiver designs.Several groups have developed through-metal ultrasonic communication systems in the recent past, though none has produced a mathematical model that accurately describes the phenomena found within the channel. The channel model developed in this thesis can be used at several stages of the transceiver design process, from transducer selection through channel equalizer design and ultimately system performance simulation.Using this channel model, we go on to develop and test several ultrasonic throughmetal transceiver designs. Ultrasonic through-metal communication systems are finding use in a wide variety of applications. Some require high throughput, while others require low power consumption. The motivation for developing several designs { ranging from low complexity, low power to high complexity, high throughput { is so that the best design can be matched to each application.After these transceiver designs are developed, we present an analysis of their computational requirements so that the most appropriate transceiver can be chosen for a given application.Ph.D., Electrical Engineering -- Drexel University, 201
Detecting a sequential signal signature in the radio frequency spectrum
No full textWe outline algorithms and methods that are useful in identifying and detecting the presence of a particular signal of interest in a noisy Radio Frequency (RF) environment. We define a signal of interest as having particular RF features (e.g., frequency or bandwidth) dictated by the equipment and the standard used to generate the signal. We formulate the concept of a target RF communication sequence signature which defines a specific sequence of communication signals, then use this concept to explore methods of detecting sequences of signals of interest. We present a review of features that would be useful in performing signal classification, and outline a methodology for extracting a particular feature (DTMF audio) from a signal. This methodology is useful in performing feature extraction for other similar features. Finally, we demonstrate that the Profile Hidden Markov Model method would be capable of handling the classification requirements of the studied scenario.M.S., Electrical Engineering -- Drexel University, 201
Distributed Detection and Fusion in Parallel Sensor Architectures
No full textParallel distributed detection system consists of several separate sensor-detector nodes (separated spatially or by their principles of operation), each with some processing capabilities. These local sensor-detectors send some information on an observed phenomenon to a centrally located Data Fusion Center for aggregation and decision making. Often, the local sensors use electro-mechanical, optical or RF modalities and are known as ``hard'' sensors. For such data sources, the sensor observations have structure and often some tractable statistical distributions which help in weighing their contribution to an integrated global decision. In a distributed detection environment, we often also have ``humans in the loop.''. Humans provide their subjective opinions on these phenomena. These opinions are labeled ``soft'' data. It is of interest to integrate "soft'' decisions, mostly assessments provided by humans, with data from the "hard" sensors, in order to improve global decision reliability. Several techniques were developed to combine data from traditional hard sensors, and a body of work was also created about integration of "soft'' data. However relatively little work was done on combining hard and soft data and decisions in an integrated environment. Our work investigates both "hard'' and "hard/soft'' fusion schemes, and proposes data integration architectures to facilitate heterogeneous sensor data fusion. In the context of "hard'' fusion, one of the contributions of this thesis is an algorithm that provides a globally optimum solution for local detector (hard sensor) design that satisfies a Neyman-Pearson criterion (maximal probability of detection under a fixed upper bound on the global false alarm rate) at the fusion center. Furthermore, the thesis also delves into application of distributed detection techniques in both parallel and sequential frameworks. Specifically, we apply parallel detection and fusion schemes to the problem of real time computer user authentication and sequential Kalman filtering for real time hypoxia detection. In the context of "hard/soft'' fusion, we propose a new Dempster-Shafer evidence theory based approach to facilitate heterogeneous sensor data fusion. Application of the framework to a number of simulated example scenarios showcases the wide range of applicability of the developed approach. We also propose and develop a hierarchical evidence tree based architecture for representing nested human opinions. The proposed framework is versatile enough to deal with both hard and soft source data using the evidence theory framework, it can handle uncertainty as well as data aggregation.Ph.D., Electrical Engineering -- Drexel University, 201
Deployment and replenishment of sensors in wireless monitoring networks
No full textRecent advances in Micro Electro-Mechanical Systems (MEMS) technology, including MEMS sensors, have allowed small, inexpensive, energy-efficient, and reliable sensors with wireless networking capabilities to become a reality. The continuing development of these technologies has given rise to increased interest in the concept of wireless sensor networks (WSN), which have been the subject of extensive research in recent years. A wireless sensor network is composed of a large number of sensor nodes, each consisting of sensing, data processing, and communication components that are deployed onto a region of interest and form a network to directly sense and report on physical phenomena. The goal of a monitoring wireless sensor network is to gather sensor data from a specified region and relay this information to a designated base station (BS).In this thesis, we focus on the problem of deploying and replenishing wireless sensor nodes onto an area such that a given mission lifetime is met subject to constraints on cost, connectivity, coverage, and capacity. The major contributions of this work are (1) a technique for differential deployment (meaning that nodes are deployed with different densities depending on their distance from the base station) in a clustered architecture that extends lifetime beyond lifetime experienced with a uniform deployment (as well as other existing differential techniques); (2) a characterization of the energy consumption in a clustered network and the energy remaining after network failure; (3) a strategy for replenishing nodes consisting of determining the optimal order size and the allocation over the deployment region. The deployment and replenishment strategies are developed through analysis of the design constraints imposed by typical sensor nodes. The result is a set of algorithms that provide differential deployment densities for nodes (clusterhead and nonclusterhead) that maximize network lifetime and minimize wasted energy, and, in cases where a single deployment is not feasible, differential densities for the optimal replenishment order sizes that minimizes deployment costs.M.S., Electrical Engineering -- Drexel University, 200
Bit-error-rate and capacity estimation in wireless networks
No full textIn this study we examine a scenario where communication over a wireless channel degrades causing applications communicating over this channel to perform poorly. Dynamically adapting protocols, such as forward-error correction (FEC), can mitigate the e ect of the link degradation. To drive these protocols we need to know the current bit-error rate (BER) and capacity of the wireless channel.In some scenarios it is not possible to gain direct access to the wireless radios and calculate the BER and capacity directly. We present a solution that can estimate the BER and capacities of these links to feed into forward error correction (FEC) or other modules using only packet information. We consider, an example scenario where it is not possible for applications to gain direct access to the wireless radios, what happens when an airborne platform communicates with a ground station.Finally we discuss the control problem that arises when multiple applications share the constrained network resource. When FEC, or another adaptive protocol, is employed, it can increase the performance of a single application, but may lower the aggregate performance of all applications sharing the constrained network link. In some scenarios applications need to be controlled (by activating the adaptive protocols) or suspended (by stopping communication of that application) so that the remaining applications can perform adequately. We first present the control problem of managing the throughput of a number of applications. We describe the Channel State (CS) algorithm, which determines how to allocate the available resources by determining which applications are to be to controlled or suspend. Then we demonstrate the operation of the CS algorithm in two scenarios, first in the presence of decreasing capacity and then in the presence of increasing BER.Ph.D., Electrical Engineering -- Drexel University, 201
Practical applications and properties of the Exponentially Modified Gaussian (EMG) distribution
No full textThe exponentially modified Gaussian (EMG) probability distribution is de ned as the convolution of an exponential distribution and a Gaussian distribution which are independent of each other. Using a reparameterized form of the EMG cumulative distribution function (cdf) several properties of the EMG distribution are derived. These properties are used to test whether the distribution of the perfect match (PM) probes from five Affymetrix microarrays follows an EMG distribution and to create a new parameter estimation method. A commonly used method for preprocessing Affymetrix microarray data, known as the robust multi-array average (RMA), assumes that the distribution of the PM probes at least approximately follows an EMG distribution. Using the results derived in this thesis it is found that the EMG distribution is not a good t for these sample data based on differences in the right tail of the sample distribution. A new distribution that is very dissimilar to the right tail of an EMG distribution is derived that more accurately fits the right tail of the sample data. From the properties of the EMG distribution derived in this thesis it is further shown that a new parameter estimation method can be created. This new parameter estimation method is compared against two other methods from the literature namely the method of moments and the Silver method (2009). From a theoretical perspective, the new parameter estimation method has the advantage that it is proven to be consistent and to always return valid parameter estimates (such as the constraint that the variance be positive). Neither the Silver method nor the method of moments has both of these qualities. All three methods were compared on the same syntheticdata generated from EMG distributions and it was found that the performance of each method depended on the "shape of the EMG distribution. It was also found that the Silver method appears to not be consistent for EMG distributions that are too "close to being a Gaussian distribution.Ph.D., Electrical Engineering -- Drexel University, 3/2
- …
