Spelling suggestions: "subject:"simplex"" "subject:"implex""
401 |
Data Mining with Newton's Method.Cloyd, James Dale 01 December 2002 (has links) (PDF)
Capable and well-organized data mining algorithms are essential and fundamental to helpful, useful, and successful knowledge discovery in databases. We discuss several data mining algorithms including genetic algorithms (GAs). In addition, we propose a modified multivariate Newton's method (NM) approach to data mining of technical data. Several strategies are employed to stabilize Newton's method to pathological function behavior. NM is compared to GAs and to the simplex evolutionary operation algorithm (EVOP). We find that GAs, NM, and EVOP all perform efficiently for well-behaved global optimization functions with NM providing an exponential improvement in convergence rate. For local optimization problems, we find that GAs and EVOP do not provide the desired convergence rate, accuracy, or precision compared to NM for technical data. We find that GAs are favored for their simplicity while NM would be favored for its performance.
|
402 |
Practical Implementations Of The Active Set Method For Support Vector Machine Training With Semi-definite KernelsSentelle, Christopher 01 January 2014 (has links)
The Support Vector Machine (SVM) is a popular binary classification model due to its superior generalization performance, relative ease-of-use, and applicability of kernel methods. SVM training entails solving an associated quadratic programming (QP) that presents significant challenges in terms of speed and memory constraints for very large datasets; therefore, research on numerical optimization techniques tailored to SVM training is vast. Slow training times are especially of concern when one considers that re-training is often necessary at several values of the models regularization parameter, C, as well as associated kernel parameters. The active set method is suitable for solving SVM problem and is in general ideal when the Hessian is dense and the solution is sparse–the case for the `1-loss SVM formulation. There has recently been renewed interest in the active set method as a technique for exploring the entire SVM regularization path, which has been shown to solve the SVM solution at all points along the regularization path (all values of C) in not much more time than it takes, on average, to perform training at a single value of C with traditional methods. Unfortunately, the majority of active set implementations used for SVM training require positive definite kernels, and those implementations that do allow semi-definite kernels tend to be complex and can exhibit instability and, worse, lack of convergence. This severely limits applicability since it precludes the use of the linear kernel, can be an issue when duplicate data points exist, and doesn’t allow use of low-rank kernel approximations to improve tractability for large datasets. The difficulty, in the case of a semi-definite kernel, arises when a particular active set results in a singular KKT matrix (or the equality-constrained problem formed using the active set is semidefinite). Typically this is handled by explicitly detecting the rank of the KKT matrix. Unfortunately, this adds significant complexity to the implementation; and, if care is not taken, numerical iii instability, or worse, failure to converge can result. This research shows that the singular KKT system can be avoided altogether with simple modifications to the active set method. The result is a practical, easy to implement active set method that does not need to explicitly detect the rank of the KKT matrix nor modify factorization or solution methods based upon the rank. Methods are given for both conventional SVM training as well as for computing the regularization path that are simple and numerically stable. First, an efficient revised simplex method is efficiently implemented for SVM training (SVM-RSQP) with semi-definite kernels and shown to out-perform competing active set implementations for SVM training in terms of training time as well as shown to perform on-par with state-of-the-art SVM training algorithms such as SMO and SVMLight. Next, a new regularization path-following algorithm for semi-definite kernels (Simple SVMPath) is shown to be orders of magnitude faster, more accurate, and significantly less complex than competing methods and does not require the use of external solvers. Theoretical analysis reveals new insights into the nature of the path-following algorithms. Finally, a method is given for computing the approximate regularization path and approximate kernel path using the warm-start capability of the proposed revised simplex method (SVM-RSQP) and shown to provide significant, orders of magnitude, speed-ups relative to the traditional grid search where re-training is performed at each parameter value. Surprisingly, it also shown that even when the solution for the entire path is not desired, computing the approximate path can be seen as a speed-up mechanism for obtaining the solution at a single value. New insights are given concerning the limiting behaviors of the regularization and kernel path as well as the use of low-rank kernel approximations. Read more
|
403 |
Implication de la protéine Mitochondriale UCP2 dans la réponse immunitaire /cLaurie RougerRouger, Laurie 13 April 2018 (has links)
La protéine découplante mitochondriale UCP2 est exprimée basalement dans différents tissus et le rôle de régulateur négatif de la production des espèces actives oxygénées (ROS) mitochondriales a été avancé. Les travaux présentés se sont intéressés à l'implication potentielle d'UCP2 dans le déroulement de la réponse immunitaire. Une induction de l'expression d'UCP2 est démontrée dans un modèle murin d'encéphalite par le virus Herpès simplex de type 1 (HSV-1) tout d'abord, puis dans un modèle parasitaire de leishmaniose causée par Leishmania donovani. Cette expression d'UCP2 a été observée dans les cellules immunitaires, colocalisée avec différents marqueurs inflammatoires exprimés à des niveaux élevés. Le modèle d'infection virale a montré un délai entre la mise en place de la réponse immunitaire innée et l'apparition d'UCP2 dans le cerveau, impliquant UCP2 dans les étapes tardives de la défense antivirale. A cette étape, la réplication du virus et la neuroinflammation sont au maximum, expliquant la susceptibilité des souris déficientes en UCP2. De plus, l'invalidation des gènes codant pour TNF-a et/ou IL-ip sensibilise au virus HSV-1 les souris naturellement résistantes et modifie le profil d'expression d'UCP2. Dans le modèle parasitaire, l'invalidation du gène codant UCP2 n'a pas permis de diminuer la croissance des parasites dans la première phase de la leishmaniose. Ainsi la protéine UCP2 ne serait pas impliquée dans les étapes précoces de mise en place de la défense antiparasitaire; les tactiques des parasites pour inhiber la réponse de l'hôte pourraient expliquer l'absence de différences nettes liées à la perte d'UCP2. La dernière étude portait sur la chronologie d'apparition d'UCP2 par rapport aux marqueurs neuroinflammatoires suite à l'injection intrastriatale de LPS. Dans ce modèle, une forte expression transitoire de TNF-a, bcBa et TLR2 est repérée dans la zone adjacente au site d'injection dès 6 h suivant l'injection puis diminue. UCP2 est seulement observable 48 h post injection dans les mêmes régions cérébrales. En conclusion, nous avons démontré dans 3 types de stimulations infectieuses de la réponse immunitaire que l'apparition d'UCP2 était liée avec un délai à l'expression de facteurs inflammatoires et UCP2 n'a semblée être qu'un marqueur. En effet, la fonction antioxydante tenue par UCP2 dans les étapes tardives de l'immunité semble modeste à la vue des résultats des souris déficientes en UCP2, mais ce rôle reste encore à préciser. / Mitochondrial uncoupling protein 2 (UCP2) is basally expressed in diverse tissues and a function of negative regulator of reactive oxygen species (ROS) production has been proposed. Studies presented in this thesis emphasize the potential role of UCP2 in immunity. Induction of UCP2 expression was demonstrated first in a murine model of encephalitis caused by Herpes simplex virus type 1 (HSV-1), then in a parasitic leishmaniasis model. UCP2 mRNA expression was observed in immune cells, colocalized with different inflammatory factors at high levels. Models of viral infection demonstrated that the transcriptional activation of UCP2 was delayed compared with the inflammatory response, involving UCP2 in the later stages of the antiviral response. At those later steps, viral replication and neuroinflammation were maximal, which could explain susceptibility in UCP2-deficient mice. Moreover, invalidation of the genes encoding TNFa and/or IL1B in resistant mice allowed HSV-1 for replicating in neurons and modified the distribution pattern of UCP2m RNA in the brain. In parasitic models, growth of parasites in UCP2- deficient mice was not impaired in the first phase of pathogenesis. Therefore UCP2 did not appear to be implicated in early steps of the antiparasitic response; parasites could develop strategies for inhibiting the host response, which would explain the similarity in the immune responses between UCP2-deficient and wild type mice. The last study addressed the chronology of UCP2 expression in relation with neuroinflammation following intrastriatal injection of LPS. In this model, strong and transitory expressions of TNFa, IkB and TLR2 were shown in ipsilateral region at 6h after injection. UCP2 was only expressed 48h post injection in the same cerebral region. In conclusion, we have demonstrated, in three models of infection triggering an immune response that the induction of UCP2 expression occurs in a coordinated but delayed manner with that of common inflammatory factors. UCP2 appears to be a marker of the severity of the immune response. It does not appear, based on the results obtained in UCP2-deficient mice, to play an essential role in the models of infection that we used. The role of UCP2 in those models remains to be folly elucidated. Read more
|
404 |
Indole-3-Carbinol Inhibition of Herpes Simplex Virus ReplicationStoner, Terri Dorene 03 December 2008 (has links)
No description available.
|
405 |
Interspecific Interactions Between Native and Non-Native Forest Floor Detritivores and Temperature: Implications for Ecosystem FunctioningMoore, Eric A. 26 July 2017 (has links)
No description available.
|
406 |
Origin Destination Problem for Traffic ControlFransholm, Elin, Hallberg, Alexander January 2024 (has links)
A typical problem in traffic control is the steering over a network of vehicles with different origins and destinations. In this report this scenario is formulated as a multi-commodity network flow problem, a linear programming problem whose objective is to transport, with minimum cost, different commodities from their respective sources to their sinks through a network, while respecting the capacity constraints of the roads. The dynamic network flow formulation of the problem is also presented, extending the network over time to incorporate the temporal dimension. Different algorithms for solving the multi-commodity network flow problem are examined. First, the simplex method, more precisely its revised version, is considered, and then the Dantzig-Wolfe decomposition is illustrated, an optimization algorithm which exploits specific block structures in the constraints. These methods are applied using state-of-the-art linear programming solvers and evaluated with a simulation based on the road network in central Stockholm. The results show that both methods allow for solving the traffic flow problem, with limitations given by the specifics of the solvers and by the space and time discretization of the problem. In particular, the revised simplex algorithm results the faster method. Read more
|
407 |
DESENVOLVIMENTO DE BEBIDA MISTA DE SORO DE LEITE COM TEOR DE LACTOSE REDUZIDO E EXTRATO HIDROSSOLÚVEL DE SOJAGregianin, Dóris de Oliveira 28 August 2014 (has links)
Made available in DSpace on 2017-07-21T18:53:23Z (GMT). No. of bitstreams: 1
Doris Oliveira Gregianin.pdf: 1718205 bytes, checksum: 59a8cb966255e87e0fa3f347f2a191cb (MD5)
Previous issue date: 2014-08-28 / The increased incidence of cardiovascular disease and various cancers have been the cause of searching for changing habits in a large portion of the population, seeking the physical activity in conjunction with the consumption of healthier foods with higher nutritional content. Considering the high biological value of the proteins of sweet whey and the recognized effect in reducing risk of cardiovascular disease of soy protein, the aim of this study was to develop a mixed chocolate-soluble soybean extract dissolved in sweet whey. The different proportions of carrageenan stabilizers, pectin and tapioca starch were defined by the experimental design of mixtures simplex centroid. By optimizing the response surface was definitely the best formulation in relation to the physic-chemical and sensory parameters. The simultaneous optimization of response variables
indicated that the optimal formulation should contain 46% carrageenan and 54% pectin and 0% starch, with a desirability index of 0.929. The formulation indicated
for the desirability function was made with water soluble soy extract dissolved in sweet whey and; on sweet whey with reduced lactose content. Serum was hydrolyzed by the action of lactase from Kluyveromyces lactis yeast enzyme at a concentration of 2500 NLU / L in reaction 4 hr at 37 ° C. The sensory attributes were evaluated by nine points hedonic scale and attitude scale in purchase intent. For overall impression, the formulations presented average value of 7.11 ±1.09, while purchase intent the average value was 3.99 ± 1.28, and the multiple
regression models proved to be predictive. The most widely accepted formulations were those who had lower viscosity and instrumental containing pectin in its constitution. The results obtained indicate that the combination of sweet whey and soy aqueous extract is a viable option for the preparation of a chocolate beverage. It was also observed that the use of cassava starch as a
thickener does not interfere negatively on the rheological and sensory characteristics and; that the combined carrageenan and pectin concentrations defined predictive model ensures greater acceptability of the product. / O aumento da incidência de doenças cardiovasculares e diversos tipos de câncer tem sido a causa da procura por mudanças de hábitos em grande parte da população, buscando-se a prática de atividades físicas em conjunto com o consumo de alimentos mais saudáveis, com maior valor nutricional. Considerando o alto valor biológico das proteínas do soro de leite doce e o reconhecido efeito na diminuição de risco de doenças cardiovasculares da
proteína de soja, o objetivo deste trabalho foi desenvolver um achocolatado misto de extrato hidrossolúvel de soja dissolvido em soro de leite doce. As
diferentes proporções dos estabilizantes carragena, pectina e amido de mandioca foram definidas pelo planejamento experimental de misturas simplex centróide. Através da otimização da Superfície de Resposta foi definida a melhor formulação em relação aos parâmetros físico-químicos e sensoriais. A otimização simultânea das variáveis de resposta indicou que a formulação ideal
deveria conter 46% de carragena e 54% de pectina e 0% de amido, com um índice de desejabilidade de 0,929. A formulação indicada pela função de desejabilidade foi elaborada com extrato hidrossolúvel de soja dissolvido em
soro de leite doce e; em soro doce com teor de lactose reduzido. O soro foi hidrolisado pela ação da enzima lactase proveniente da levedura Kluyveromyces lactis, na concentração de 2.500 NLU/L numa reação de 4 horas à temperatura de 37°C. Os atributos sensoriais foram avaliados pela escala hedônica de nove pontos e escala de atitude na intenção de compra. Para impressão global as formulações apresentaram valor médio 7,11 ±1,09, enquanto que para intenção de compra o valor médio foi de 3,99 ±1,28, sendo que os modelos de regressão múltipla mostraram-se preditivos. As formulações mais aceitas foram aquelas que apresentaram menor viscosidade instrumental e que continham pectina em sua constituição. Os resultados obtidos indicam que a
combinação do soro doce de leite e extrato aquoso de soja é uma opção viável para a elaboração de uma bebida achocolatada. Também foi possível observar que o uso do amido de mandioca como espessante não interfere negativamente nas características reológicas e sensoriais do produto e; que a carragena e pectinas combinadas, nas concentrações definidas pelo modelo preditivo,
garantem maior aceitabilidade ao produto. Read more
|
408 |
Contribución a los métodos de optimización basados en procesos naturales y su aplicación a la medida de antenas en campo próximoPérez López, Jesús Ramón 16 December 2005 (has links)
Durante la última década, los métodos de optimización heurísticos basados en imitar a nivel computacional procesos naturales, biológicos, sociales o culturales, han despertado el interés de la comunidad científica debido a su capacidad para explorar eficientemente espacios de soluciones multimodales y multidimensionales. En este ámbito, esta tesis aborda el desarrollo, análisis y puesta a punto de diferentes métodos de optimización tradicionales y heurísticos. En concreto, se considera un método de búsqueda local basado en símplex y varios métodos heurísticos, tales como el recocido simulado, los algoritmos genéticos y la optimización con enjambre de partículas. Para estos dos últimos algoritmos se investigan diferentes esquemas con el objetivo de superar las limitaciones propias de los esquemas clásicos.La puesta a punto de los diferentes métodos de optimización se realiza considerando como problema de referencia la caracterización de la radiación de antenas a partir de medidas en campo próximo sobre geometría plana, utilizando un método de transformación de campo cercano a campo lejano basado en corrientes equivalentes. Para cada método de optimización se incluye un análisis paramétrico, en los casos en los que se ha considerado necesario, así como resultados de transformación de campo teóricos obtenidos para diferentes antenas de apertura y antenas de bocina piramidal. Los resultados de un estudio comparativo, realizado utilizando fuentes teóricas y medidas, demuestran la utilidad del método y permiten concluir que la optimización con enjambre de partículas es el algoritmo que proporciona las mejores prestaciones para esta aplicación.Los métodos de optimización desarrollados e investigados en este trabajo han sido también aplicados a otros problemas como son la síntesis de agrupaciones lineales o el modelado de fuente en aplicaciones de compatibilidad electromagnética. / For the last decade, heuristic optimization methods based on imitating natural, biological, social or cultural processes in a computational way have aroused great interest among the scientific community, due to its ability to explore efficiently multimodal and high-dimension solution spaces. On this basis, this thesis tackles the development, analysis and tuning of different traditional and heuristic optimization methods. In short, a local search method based on simplex and several heuristic methods, such as simulated annealing, genetic algorithms and particle swarm optimization are considered. For these last two algorithms different schemes are investigated so as to overcome the typical limitations of classical schemes.The tuning of the optimization methods is carried out considering as a reference problem the antenna radiation characterization from near-field measurements over a planar geometry, using a near-field to far-field transformation method based on equivalent currents. A parametric analysis is included for each optimization method, in those cases in which it has been considered necessary, as well as theoretical field transformation results obtained with aperture and pyramidal horn antennas. Results of a comparative study, carried out using theoretical sources and measurements, demonstrate the usefulness of the method and make it possible to conclude that the particle swarm optimization is the algorithm that provides the best performance for this application.The optimization methods developed and investigated in this work have also been applied to other problems, such as the synthesis of linear arrays or the source modelling in electromagnetic compatibility applications. Read more
|
409 |
Vergleichende Analysen zur Replikation und zum intraaxonalen Transport des Pseudorabiesvirus und des Herpes Simplex Virus Typ 1 in primären RattenneuronenNegatsch, Alexandra 28 September 2015 (has links) (PDF)
Nach dem Eintritt in den Wirtsorganismus und initialer Replikation infizieren Alphaherpesviren Neuronen zur weiteren Ausbreitung im Nervensystem und zur Etablierung einer Latenz. Dazu werden die Viruspartikel innerhalb der Axone retrograd von der Peripherie zum neuronalen Zellkörper transportiert. Die umgekehrte Richtung beschreibt den Weg des anterograden Transports vom Zellkörper zur Synapse für weitere Infektionen von Neuronen höherer Ordnung oder zurück zur Peripherie. Der retrograde intraaxonale Transport ist gut untersucht. Dagegen wird über den anterograden Transport kontrovers diskutiert. Zwei verschiedene Transportmodelle werden vermutet. Das „Married Model“ postuliert, dass umhüllte Virionen innerhalb von Vesikeln entlang des Axons transportiert werden. Die Freisetzung der Partikel erfolgt an der jeweiligen Synapse durch Endocytose. Das „Subassembly Model“ geht dagegen davon aus, dass einzelne Virusstrukurkomponenten (Nukleokapsid, Hülle) entlang des Axons transportiert werden. Der Zusammenbau und die Freisetzung erfolgt am Axonterminus bzw. an der Synapse (in vivo) oder am Wachstumskegel (in vitro) oder an speziellen Auftreibungen des Axons, den sogenannten Varicosities. Nach Infektion eines neuronalen Explantatsystems mit dem Pseudorabiesvirus (PrV) konnten ultrastrukturell umhüllte Virionen in Vesikeln detektiert werden und so der Nachweis der Gültigkeit des „Married Model“ als vorherrschendes Transportmodell geführt werden. Dagegen ist die Situation beim prototypischen Alphaherpesvirus, dem Herpes Simplex Virus Typ 1 (HSV-1), weiterhin ungeklärt. Aufgrund der zahlreichen unterschiedlichen Analysemethoden und -systeme war ein direkter Vergleich der beiden Viren bislang nicht möglich. Daher sollte in dieser Arbeit ein standardisiertes neuronales Kultursystem genutzt werden, um vier verschiedene HSV-1 Stämme im Vergleich zu PrV zu untersuchen. Für die Infektionen wurden sowohl Neuronen aus dem oberen Cervikalganglion als auch aus Spinalganglien genutzt. So konnte gezeigt werden, dass in Neuronen, welche mit den HSV-1 Stämmen HFEM, 17+ und SC16 infiziert waren ca. 75% als umhüllte Virionen in Vesikeln und ca. 25% als nackte Kapside vorlagen. Ingesamt war die Anzahl der Viruspartikel in HSV-1 infizierten Neuronen signifikant geringer als in PrV infizierten Kulturen. Überraschenderweise zeigten mit HSV-1 KOS infizierte Neuronen ein reverses Bild. Hier lagen nur 25% der Viruspartikel als umhüllte Virionen in Vesikeln vor, während 75% als nackte Kapside detektiert wurden. Dieser unerwartete Phänotyp sollte auf molekularbiologischer Ebene genauer untersucht werden. Dabei wurde auf die Genregion von US9 fokussiert. Das von US9 codierte Membranprotein spielt eine wichtige Rolle während des Zusammenbaus der Virionen und bei anschließenden axonalen anterograden Transportvorgängen. In dieser Arbeit konnte gezeigt werden, dass das HSV-1 KOS Genom durch verschiedene Basenaustausche an der vorhergesagten TATA-Box von US9 eine Mutation aufweist. Zusätzlich trägt das offene Leseraster durch eine weitere Mutation ein vorzeitiges Stopcodon auf und wird dadurch auf 58 Kodons reduziert, im Gegensatz zu anderen HSV-1 Stämmen, wo es 91 Kodons umfasst. Die Mutation an der TATA-Box verändert auch das ursprüngliche Stopcodon vom US8a Gen, was zur einer Verlängerung von ursprünglich 161 zu 191 Kodons führt. In Northern Blot Analysen konnte eine reduzierte Transkription von US9 in HSV-1 KOS infizierten Zellen detektiert werden. In HSV-1 KOS infizierten Zellen konnten mittels eines spezifischen Antiserums gegen US9 im Western Blot kein Genprodukt nachgewiesen werden. Auch Immunfluoreszenzanalysen zeigten, dass das abgeleitete verkürzte Protein offenbar nicht stabil exprimiert wird. Dagegen konnten Western Blot Analysen die Vergrößerung des pUS8a bestätigen. Der beobachtete auffällige intraaxonale Phänotyp könnte somit durch die Mutation des US9 Protein erklärt werden. Zusammenfassend wurde in dieser Arbeit gezeigt, dass auch bei HSV-1 vorwiegend das „Married Model“ für den anterograden intraaxonalen Transportweg bevorzugt wird und somit beide Alphaherpesviren, HSV-1 und PrV, denselben Transportweg nutzen. Read more
|
410 |
Επεξεργασία οφθαλμολογικών εικόνων για μέτρηση διαμέτρων αγγείωνΒλαχοκώστα, Αλεξάνδρα 27 August 2008 (has links)
Σκοπός της παρούσας Διπλωματικής Εργασίας είναι η ανάπτυξη συγκεκριμένης μεθοδολογίας και αλγορίθμων ψηφιακής επεξεργασίας εικόνων για την αυτόματη εκτίμηση των διαμέτρων αγγείων σε οφθαλμολογικές εικόνες. Η συγκεκριμένη μέτρηση της διαμέτρου των αγγείων διαδραματίζει σημαντικό ρόλο στην έγκαιρη διάγνωση παθήσεων καθώς έχει αποδειχθεί ότι υπάρχει συσχετισμός μεταξύ των μεταβολών των τιμών των εν λόγω διαμέτρων και της εμφάνισης αλλοιώσεων στον αμφιβληστροειδή.
Στα πλαίσια της εργασίας, υλοποιήθηκαν δύο μεθοδολογίες για τον υπολογισμό των διαμέτρων αγγείων οφθαλμολογικών εικόνων, οι οποίες συλλέγονται με χρήση κάμερας πυθμένα (fundus camera). Η πρώτη μεθοδολογία στηρίζεται στην εύρεση των σημείων που αποτελούν τους κεντρικούς άξονες των υπό εξέταση αγγείων με χρήση διαφορικού λογισμού. Ακολούθως, σε κάθε σημείο που ανήκει σε κεντρικό άξονα αγγείου, υπολογίζονται οι παράμετροι μιας συνάρτησης. Η εν λόγω συνάρτηση περιγράφει βέλτιστα τα επίπεδα φωτεινότητας της εικόνας κατά μήκος του ευθύγραμμου τμήματος που διέρχεται από το σημείο και είναι κάθετο στο αγγείο. H εύρεση των παραμέτρων της συνάρτησης πραγματοποιείται με χρήση τεχνικών βελτιστοποίησης. Το τελικό βήμα της μεθοδολογίας είναι η εκτίμηση της διαμέτρου των αγγείων από τις τιμές των παραμέτρων που έχουν υπολογιστεί.
Η δεύτερη μεθοδολογία στηρίζεται στον αλγόριθμο που προτείνει ο P.H. Gregson. Αρχικά, πραγματοποιείται κατάτμηση της εικόνας με κατωφλίωση και εφαρμόζονται μορφολογικοί τελεστές συστολής και διαστολής στην εικόνα. Στη συνέχεια, εφαρμόζεται ο αλγόριθμος λέπτυνσης (thinning algorithm) με σκοπό την εύρεση των κεντρικών αξόνων των αγγείων και τέλος εκτιμάται η διάμετρος σε κάθε σημείο του κεντρικού άξονα με χρήση των επιπέδων του γκρίζου των εικονοστοιχείων που κείνται στην ευθεία που είναι κάθετη στο αγγείο σε κάθε σημείο του. / The scope of this Thesis is the development of a methodology and advance image processing techniques in order to automatically estimate vessel diameters in ophthalmological images. Motivation for the thesis is the fact that the measurement of vessel diameter plays significant role in the seasonable diagnosis of vascular disorders, as it is believed to be a relation between the variation in diameters and the detection of retinal disorders.
In this thesis, two methodologies are developed in order to be applied in ophthalmological images that are collected by using a fundus camera. The first methodology is based on the detection of the pixels that constitute the centerlines of vessels, by using differential calculus. Specifically, at each pixel that belongs to a centerline of vessel, the parameters of a specific function are calculated. This function describes as accurately as possible the intensity levels along the segment that passes through the specific pixel and is perpendicular to the vessel. The parameters of this function are estimated using optimization techniques. The final step of the methodology is the assessment of the diameters of vessels using the values of the parameters.
The second methodology is based on the algorithm that P.H.Gregson has proposed. At first, the vessels are detected by tresholding and a morphological closing algorithm is applied. Then, a thinning algorithm is used in order to detect the pixels that constitute the centerlines of the vessels and ultimately the diameter at each pixel of the centerlines is assessed using the gray levels of the pixels that constitute the segment that is perpendicular to the vessel at each specific pixel. Read more
|
Page generated in 0.0396 seconds