Microstructural segmentation of 3D images of alsphalt specimen using Matlab engine

En este proyecto se ha desarrollado un código de MATLAB para el procesamiento de imágenes tomográficas 3D, de muestras de asfalto de carreteras en Polonia. Estas imágenes en 3D han sido tomadas por un equipo de investigación de la Universidad Tecnológica de Lodz (LUT). El objetivo de este proyecto es crear una herramienta que se pueda utilizar para estudiar las diferentes muestras de asfalto 3D y pueda servir para estudiar las pruebas de estrés que experimentan las muestras en el laboratorio. Con el objetivo final de encontrar soluciones a la degradación sufrida en las carreteras de Polonia, debido a diferentes causas, como son las condiciones meteorológicas. La degradación de las carreteras es un tema que se ha investigado desde hace muchos años, debido a la fuerte degradación causada por diferentes factores como son climáticos, la falta de mantenimiento o el tráfico excesivo en algunos casos. Es en Polonia, donde estos tres factores hacen que la composición de muchas carreteras se degrade rápidamente, sobre todo debido a las condiciones meteorológicas sufridas a lo largo del año, con temperaturas que van desde 30° C en verano a -20° C en invierno. Esto hace que la composición de las carreteras sufra mucho y el asfalto se levante, lo que aumenta los costos de mantenimiento y los accidentes de carretera. Este proyecto parte de la base de investigación que se lleva a cabo en la LUT, tratando de mejorar el análisis de las muestras de asfalto, por lo que se realizarán las pruebas de estrés y encontrar soluciones para mejorar el asfalto en las carreteras polacas. Esto disminuiría notablemente el costo de mantenimiento. A pesar de no entrar en aspectos muy técnicos sobre el asfalto y su composición, se ha necesitado realizar un estudio profundo sobre todas sus características, para crear un código capaz de obtener los mejores resultados. Por estas razones, se ha desarrollado en Matlab, los algoritmos que permiten el estudio de los especímenes 3D de asfalto. Se ha utilizado este software, ya que Matlab es una poderosa herramienta matemática que permite operar con matrices para realización de operaciones rápidamente, permitiendo desarrollar un código específico para el tratamiento y procesamiento de imágenes en 3D. Gracias a esta herramienta, estos algoritmos realizan procesos tales como, la segmentación de la imagen 3D, pre y post procesamiento de la imagen, filtrado o todo tipo de análisis microestructural de las muestras de asfalto que se están estudiando. El código presentado para la segmentación de las muestras de asfalto 3D es menos complejo en su diseño y desarrollo, debido a las herramientas de procesamiento de imágenes que incluye Matlab, que facilitan significativamente la tarea de programación, así como el método de segmentación utilizado. Respecto al código, este ha sido diseñado teniendo en cuenta el objetivo de facilitar el trabajo de análisis y estudio de las imágenes en 3D de las muestras de asfalto. Por lo tanto, el principal objetivo es el de crear una herramienta para el estudio de este código, por ello fue desarrollado para que pueda ser integrado en un entorno visual, de manera que sea más fácil y simple su utilización. Ese es el motivo por el cual todos estos algoritmos y funciones, que ha sido desarrolladas, se integrarán en una herramienta visual que se ha desarrollado con el GUIDE de Matlab. Esta herramienta ha sido creada en colaboración con Jorge Vega, y fue desarrollada en su proyecto final de carrera, cuyo título es: Segmentación microestructural de Imágenes en 3D de la muestra de asfalto utilizando Matlab. En esta herramienta se ha utilizado todo las funciones programadas en este proyecto, y tiene el objetivo de desarrollar una herramienta que permita crear un entorno gráfico intuitivo y de fácil uso para el estudio de las muestras de 3D de asfalto. Este proyecto se ha dividido en 4 capítulos, en un primer lugar estará la introducción, donde se presentarán los aspectos más importante que se va a componer el proyecto. En el segundo capítulo se presentarán todos los datos técnicos que se han tenido que estudiar para desarrollar la herramienta, entre los que cabe los tres temas más importantes que se han estudiado en este proyecto: materiales asfálticos, los principios de la tomografías 3D y el procesamiento de imágenes. Esta será la base para el tercer capítulo, que expondrá la metodología utilizada en la elaboración del código, con la explicación del entorno de trabajo utilizado en Matlab y todas las funciones de procesamiento de imágenes utilizadas. Además, se muestra todo el código desarrollado, así como una descripción teórica de los métodos utilizados para el pre-procesamiento y segmentación de las imagenes en 3D. En el capítulo 4, se mostrarán los resultados obtenidos en el estudio de una de las muestras de asfalto, y, finalmente, el último capítulo se basa en las conclusiones sobre el desarrollo de este proyecto. En este proyecto se ha llevado han realizado todos los puntos que se establecieron como punto de partida en el anteproyecto para crear la herramienta, a pesar de que se ha dejado para futuros proyectos nuevas posibilidades de este codigo, como por ejemplo, la detección automática de las diferentes regiones de una muestra de asfalto debido a su composición. Como se muestra en este proyecto, las técnicas de procesamiento de imágenes se utilizan cada vez más en multitud áreas, como pueden ser industriales o médicas. En consecuencia, este tipo de proyecto tiene multitud de posibilidades, y pudiendo ser la base para muchas nuevas aplicaciones que se puedan desarrollar en un futuro. Por último, se concluye que este proyecto ha contribuido a fortalecer las habilidades de programación, ampliando el conocimiento de Matlab y de la teoría de procesamiento de imágenes. Del mismo modo, este trabajo proporciona una base para el desarrollo de un proyecto más amplio cuyo alcance será una herramienta que puedas ser utilizada por el equipo de investigación de la Universidad Tecnológica de Lodz y en futuros proyectos. ABSTRACT In this project has been developed one code in MATLAB to process X-ray tomographic 3D images of asphalt specimens. These images 3D has been taken by a research team of the Lodz University of Technology (LUT). The aim of this project is to create a tool that can be used to study differents asphalt specimen and can be used to study them after stress tests undergoing the samples. With the final goal to find solutions to the degradation suffered roads in Poland due to differents causes, like weather conditions. The degradation of the roads is an issue that has been investigated many years ago, due to strong degradation suffered caused by various factors such as climate, poor maintenance or excessive traffic in some cases. It is in Poland where these three factors make the composition of many roads degrade rapidly, especially due to the weather conditions suffered along the year, with temperatures ranging from 30 o C in summer to -20 ° C in winter. This causes the roads suffers a lot and asphalt rises shortly after putting, increasing maintenance costs and road accident. This project part of the base that research is taking place at the LUT, in order to better analyze the asphalt specimens, they are tested for stress and find solutions to improve the asphalt on Polish roads. This would decrease remarkable maintenance cost. Although this project will not go into the technical aspect as asphalt and composition, but it has been required a deep study about all of its features, to create a code able to obtain the best results. For these reasons, there have been developed in Matlab, algorithms that allow the study of 3D specimens of asphalt. Matlab is a powerful mathematical tool, which allows arrays operate fastly, allowing to develop specific code for the treatment and processing of 3D images. Thus, these algorithms perform processes such as the multidimensional matrix sgementation, pre and post processing with the same filtering algorithms or microstructural analysis of asphalt specimen which being studied. All these algorithms and function that has been developed to be integrated into a visual tool which it be developed with the GUIDE of Matlab. This tool has been created in the project of Jorge Vega which name is: Microstructural segmentation of 3D images of asphalt specimen using Matlab engine. In this tool it has been used all the functions programmed in this project, and it has the aim to develop an easy and intuitive graphical environment for the study of 3D samples of asphalt. This project has been divided into 4 chapters plus the introduction, the second chapter introduces the state-of-the-art of the three of the most important topics that have been studied in this project: asphalt materials, principle of X-ray tomography and image processing. This will be the base for the third chapter, which will outline the methodology used in developing the code, explaining the working environment of Matlab and all the functions of processing images used. In addition, it will be shown all the developed code created, as well as a theoretical description of the methods used for preprocessing and 3D image segmentation. In Chapter 4 is shown the results obtained from the study of one of the specimens of asphalt, and finally the last chapter draws the conclusions regarding the development of this project.

Generación de modelos físicos personalizados y segmentación de la pared arterial y el trombo intraluminal en imágenes de aneurismas aórticos abdominales

Esta tesis doctoral está encuadrada dentro del marco general de la ingeniería biomédica aplicada al tratamiento de las enfermedades cardiovasculares, enfermedades que provocan alrededor de 1.9 millones (40%) de muertes al año en la Unión Europea. En este contexto surge el proyecto europeo SCATh-Smart Catheterization, cuyo objetivo principal es mejorar los procedimientos de cateterismo aórtico introduciendo nuevas tecnologías de planificación y navegación quirúrgica y minimizando el uso de fluoroscopía. En particular, esta tesis aborda el modelado y diagnóstico de aneurismas aórticos abdominales (AAA) y del trombo intraluminal (TIL), allí donde esté presente, así como la segmentación de estas estructuras en imágenes preoperatorias de RM. Los modelos físicos específicos del paciente, construidos a partir de imágenes médicas preoperatorias, tienen múltiples usos, que van desde la evaluación preoperatoria de estructuras anatómicas a la planificación quirúrgica para el guiado de catéteres. En el diagnóstico y tratamiento de AAA, los modelos físicos son útiles a la hora de evaluar diversas variables biomecánicas y fisiológicas de las estructuras vasculares. Existen múltiples técnicas que requieren de la generación de modelos físicos que representen la anatomía vascular. Una de las principales aplicaciones de los modelos físicos es el análisis de elementos finitos (FE). Las simulaciones de FE para AAA pueden ser específicas para el paciente y permiten modelar estados de estrés complejos, incluyendo los efectos provocados por el TIL. La aplicación de métodos numéricos de análisis tiene como requisito previo la generación de una malla computacional que representa la geometría de interés mediante un conjunto de elementos poliédricos, siendo los hexaédricos los que presentan mejores resultados. En las estructuras vasculares, generar mallas hexaédricas es un proceso especialmente exigente debido a la compleja anatomía 3D ramificada. La mayoría de los AAA se encuentran situados en la bifurcación de la arteria aorta en las arterias iliacas y es necesario modelar de manera fiel dicha bifurcación. En el caso de que la sangre se estanque en el aneurisma provocando un TIL, éste forma una estructura adyacente a la pared aórtica. De este modo, el contorno externo del TIL es el mismo que el contorno interno de la pared, por lo que las mallas resultantes deben reflejar esta particularidad, lo que se denomina como "mallas conformadas". El fin último de este trabajo es modelar las estructuras vasculares de modo que proporcionen nuevas herramientas para un mejor diagnóstico clínico, facilitando medidas de riesgo de rotura de la arteria, presión sistólica o diastólica, etc. Por tanto, el primer objetivo de esta tesis es diseñar un método novedoso y robusto para generar mallas hexaédricas tanto de la pared aórtica como del trombo. Para la identificación de estas estructuras se utilizan imágenes de resonancia magnética (RM). Deben mantenerse sus propiedades de adyacencia utilizando elementos de alta calidad, prestando especial atención al modelado de la bifurcación y a que sean adecuadas para el análisis de FE. El método tiene en cuenta la evolución de la línea central del vaso en el espacio tridimensional y genera la malla directamente a partir de las imágenes segmentadas, sin necesidad de reconstruir superficies triangulares. Con el fin de reducir la intervención del usuario en el proceso de generación de las mallas, es también objetivo de esta tesis desarrollar un método de segmentación semiautomática de las distintas estructuras de interés. Las principales contribuciones de esta tesis doctoral son: 1. El diseño, implementación y evaluación de un algoritmo de generación de mallas hexaédricas conformadas de la pared y el TIL a partir de los contornos segmentados en imágenes de RM. Se ha llevado a cabo una evaluación de calidad que determine su aplicabilidad a métodos de FE. Los resultados demuestran que el algoritmo desarrollado genera mallas conformadas de alta calidad incluso en la región de la bifurcación, que son adecuadas para su uso en métodos de análisis de FE. 2. El diseño, implementación y evaluación de un método de segmentación automático de las estructuras de interés. La luz arterial se segmenta de manera semiautomática utilizando un software disponible a partir de imágenes de RM con contraste. Los resultados de este proceso sirven de inicialización para la segmentación automática de las caras interna y externa de la pared aórtica utilizando métodos basado en modelos de textura y forma a partir de imágenes de RM sin contraste. Los resultados demuestran que el algoritmo desarrollado proporciona segmentaciones fieles de las distintas estructuras de interés. En conclusión, el trabajo realizado en esta tesis doctoral corrobora las hipótesis de investigación postuladas, y pretende servir como aportación para futuros avances en la generación de modelos físicos de geometrías biológicas. ABSTRACT The frame of this PhD Thesis is the biomedical engineering applied to the treatment of cardiovascular diseases, which cause around 1.9 million deaths per year in the European Union and suppose about 40% of deaths per year. In this context appears the European project SCATh-Smart Catheterization. The main objective of this project is creating a platform which improves the navigation of catheters in aortic catheterization minimizing the use of fluoroscopy. In the framework of this project, the specific field of this PhD Thesis is the diagnosis and modeling of abdominal aortic aneurysm (AAAs) and the intraluminal thrombus (ILT) whenever it is present. Patient-specific physical models built from preoperative imaging are becoming increasingly important in the area of minimally invasive surgery. These models can be employed for different purposes, such as the preoperatory evaluation of anatomic structures or the surgical planning for catheter guidance. In the specific case of AAA diagnosis and treatment, physical models are especially useful for evaluating pressures over vascular structures. There are multiple techniques that require the generation of physical models which represent the target anatomy. Finite element (FE) analysis is one the principal applications for physical models. FE simulations for AAA may be patient-specific and allow modeling biomechanical and physiological variables including those produced by ILT, and also the segmentation of those anatomical structures in preoperative MR images. Applying numeric methods requires the generation of a proper computational mesh. These meshes represent the patient anatomy using a set of polyhedral elements, with hexahedral elements providing better results. In the specific case of vascular structures, generating hexahedral meshes is a challenging task due to the complex 3D branching anatomy. Each patient’s aneurysm is unique, characterized by its location and shape, and must be accurately represented for subsequent analyses to be meaningful. Most AAAs are located in the region where the aorta bifurcates into the iliac arteries and it is necessary to model this bifurcation precisely and reliably. If blood stagnates in the aneurysm and forms an ILT, it exists as a conforming structure with the aortic wall, i.e. the ILT’s outer contour is the same as the wall’s inner contour. Therefore, resulting meshes must also be conforming. The main objective of this PhD Thesis is designing a novel and robust method for generating conforming hexahedral meshes for the aortic wall and the thrombus. These meshes are built using largely high-quality elements, especially at the bifurcation, that are suitable for FE analysis of tissue stresses. The method accounts for the evolution of the vessel’s centerline which may develop outside a single plane, and generates the mesh directly from segmented images without the requirement to reconstruct triangular surfaces. In order to reduce the user intervention in the mesh generation process is also a goal of this PhD. Thesis to develop a semiautomatic segmentation method for the structures of interest. The segmentation is performed from magnetic resonance image (MRI) sequences that have tuned to provide high contrast for the arterial tissue against the surrounding soft tissue, so that we determine the required information reliably. The main contributions of this PhD Thesis are: 1. The design, implementation and evaluation of an algorithm for generating hexahedral conforming meshes of the arterial wall and the ILT from the segmented contours. A quality inspection has been applied to the meshes in order to determine their suitability for FE methods. Results show that the developed algorithm generates high quality conforming hexahedral meshes even at the bifurcation region. Thus, these meshes are suitable for FE analysis. 2. The design, implementation and evaluation of a semiautomatic segmentation method for the structures of interest. The lumen is segmented in a semiautomatic way from contrast filled MRI using an available software. The results obtained from this process are used to initialize the automatic segmentation of the internal and external faces of the aortic wall. These segmentations are performed by methods based on texture and shape models from MRI with no contrast. The results show that the algorithm provides faithful segmentations of the structures of interest requiring minimal user intervention. In conclusion, the work undertaken in this PhD. Thesis verifies the investigation hypotheses. It intends to serve as basis for future physical model generation of proper biological anatomies used by numerical methods.

Classifying image analysis techniques from their output

In this paper we discuss some main image processing techniques in order to propose a classification based upon the output these methods provide. Because despite a particular image analysis technique can be supervised or unsupervised, and can allow or not the existence of fuzzy information at some stage, each technique has been usually designed to focus on a specific objective, and their outputs are in fact different according to each objective. Thus, they are in fact different methods. But due to the essential relationship between them they are quite often confused. In particular, this paper pursues a clarification of the differences between image segmentation and edge detection, among other image processing techniques.

Hough Transform recursive evaluation using Distributed Arithmetic

Paper submitted to the IFIP International Conference on Very Large Scale Integration (VLSI-SOC), Darmstadt, Germany, 2003.

Automatic texture segmentation for texture-based image retrieval

Texture-segmentation is the crucial initial step for texture-based image retrieval. Texture is the main difficulty faced to a segmentation method. Many image segmentation algorithms either can’t handle texture properly or can’t obtain texture features directly during segmentation which can be used for retrieval purpose. This paper describes an automatic texture segmentation algorithm based on a set of features derived from wavelet domain, which are effective in texture description for retrieval purpose. Simulation results show that the proposed algorithm can efficiently capture the textured regions in arbitrary images, with the features of each region extracted as well. The features of each textured region can be directly used to index image database with applications as texture-based image retrieval.

The mosaic test:benchmarking colour-based image retrieval systems using image mosaics

Evaluation and benchmarking in content-based image retrieval has always been a somewhat neglected research area, making it difficult to judge the efficacy of many presented approaches. In this paper we investigate the issue of benchmarking for colour-based image retrieval systems, which enable users to retrieve images from a database based on lowlevel colour content alone. We argue that current image retrieval evaluation methods are not suited to benchmarking colour-based image retrieval systems, due in main to not allowing users to reflect upon the suitability of retrieved images within the context of a creative project and their reliance on highly subjective ground-truths. As a solution to these issues, the research presented here introduces the Mosaic Test for evaluating colour-based image retrieval systems, in which test-users are asked to create an image mosaic of a predetermined target image, using the colour-based image retrieval system that is being evaluated. We report on our findings from a user study which suggests that the Mosaic Test overcomes the major drawbacks associated with existing image retrieval evaluation methods, by enabling users to reflect upon image selections and automatically measuring image relevance in a way that correlates with the perception of many human assessors. We therefore propose that the Mosaic Test be adopted as a standardised benchmark for evaluating and comparing colour-based image retrieval systems.

The mosaic test:measuring the effectiveness of colour-based image retrieval

A variety of content-based image retrieval systems exist which enable users to perform image retrieval based on colour content - i.e., colour-based image retrieval. For the production of media for use in television and film, colour-based image retrieval is useful for retrieving specifically coloured animations, graphics or videos from large databases (by comparing user queries to the colour content of extracted key frames). It is also useful to graphic artists creating realistic computer-generated imagery (CGI). Unfortunately, current methods for evaluating colour-based image retrieval systems have 2 major drawbacks. Firstly, the relevance of images retrieved during the task cannot be measured reliably. Secondly, existing methods do not account for the creative design activity known as reflection-in-action. Consequently, the development and application of novel and potentially more effective colour-based image retrieval approaches, better supporting the large number of users creating media for use in television and film productions, is not possible as their efficacy cannot be reliably measured and compared to existing technologies. As a solution to the problem, this paper introduces the Mosaic Test. The Mosaic Test is a user-based evaluation approach in which participants complete an image mosaic of a predetermined target image, using the colour-based image retrieval system that is being evaluated. In this paper, we introduce the Mosaic Test and report on a user evaluation. The findings of the study reveal that the Mosaic Test overcomes the 2 major drawbacks associated with existing evaluation methods and does not require expert participants. © 2012 Springer Science+Business Media, LLC.

Image Quotient Set Transforms in Segmentation Problems

Image content interpretation is much dependent on segmentations efficiency. Requirements for the image recognition applications lead to a nessesity to create models of new type, which will provide some adaptation between law-level image processing, when images are segmented into disjoint regions and features are extracted from each region, and high-level analysis, using obtained set of all features for making decisions. Such analysis requires some a priori information, measurable region properties, heuristics, and plausibility of computational inference. Sometimes to produce reliable true conclusion simultaneous processing of several partitions is desired. In this paper a set of operations with obtained image segmentation and a nested partitions metric are introduced.

Comparison of different space indexing methods for ecological evaluation of urban openspaces

Nowadays we meet many different evaluation methods regarding the ecological performance of green surfaces and parks. All these methods are extremely valuable in determining how well a green surface performs from ecological aspect and to what extent the environment were damaged if these sites would be built or would be developed any other way causing reduction of green surfaces. The goal of the article is to clarify the differences between two evaluation methods (GSI – Green Space Intensity, BARC – Biological Activity Rate Calculation) suitable for urban green infrastructure analysis and to see if any significant difference can be observed evaluating the same site by these methods. Our research sites are in Budapest and their sizes vary between 2,5-8 acres. The most important aspects of site analysis are the following: size and boundaries of the park, existence or lack of water features, the characteristics of their surfaces and the complexity of vegetation. We summarize the data of the site analysis in tables, make a summarizing diagram for visual representation and draw conclusions from the results. As a final step, we evaluate how these two evaluation systems relate to urban open space developments.

Analysis of the scale effect in different computed tomography systems on the evaluation of bone tissue parameters

Tra le patologie ossee attualmente riconosciute, l’osteoporosi ricopre il ruolo di protagonista data le sua diffusione globale e la multifattorialità delle cause che ne provocano la comparsa. Essa è caratterizzata da una diminuzione quantitativa della massa ossea e da alterazioni qualitative della micro-architettura del tessuto osseo con conseguente aumento della fragilità di quest’ultimo e relativo rischio di frattura. In campo medico-scientifico l’imaging con raggi X, in particolare quello tomografico, da decenni offre un ottimo supporto per la caratterizzazione ossea; nello specifico la microtomografia, definita attualmente come “gold-standard” data la sua elevata risoluzione spaziale, fornisce preziose indicazioni sulla struttura trabecolare e corticale del tessuto. Tuttavia la micro-CT è applicabile solo in-vitro, per cui l’obiettivo di questo lavoro di tesi è quello di verificare se e in che modo una diversa metodica di imaging, quale la cone-beam CT (applicabile invece in-vivo), possa fornire analoghi risultati, pur essendo caratterizzata da risoluzioni spaziali più basse. L’elaborazione delle immagini tomografiche, finalizzata all’analisi dei più importanti parametri morfostrutturali del tessuto osseo, prevede la segmentazione delle stesse con la definizione di una soglia ad hoc. I risultati ottenuti nel corso della tesi, svolta presso il Laboratorio di Tecnologia Medica dell’Istituto Ortopedico Rizzoli di Bologna, mostrano una buona correlazione tra le due metodiche quando si analizzano campioni definiti “ideali”, poiché caratterizzati da piccole porzioni di tessuto osseo di un solo tipo (trabecolare o corticale), incluso in PMMA, e si utilizza una soglia fissa per la segmentazione delle immagini. Diversamente, in casi “reali” (vertebre umane scansionate in aria) la stessa correlazione non è definita e in particolare è da escludere l’utilizzo di una soglia fissa per la segmentazione delle immagini.

Statistical methods for polyhedral shape classification with incomplete data - application to cryo-electron tomographic images

A certain type of bacterial inclusion, known as a bacterial microcompartment, was recently identified and imaged through cryo-electron tomography. A reconstructed 3D object from single-axis limited angle tilt-series cryo-electron tomography contains missing regions and this problem is known as the missing wedge problem. Due to missing regions on the reconstructed images, analyzing their 3D structures is a challenging problem. The existing methods overcome this problem by aligning and averaging several similar shaped objects. These schemes work well if the objects are symmetric and several objects with almost similar shapes and sizes are available. Since the bacterial inclusions studied here are not symmetric, are deformed, and show a wide range of shapes and sizes, the existing approaches are not appropriate. This research develops new statistical methods for analyzing geometric properties, such as volume, symmetry, aspect ratio, polyhedral structures etc., of these bacterial inclusions in presence of missing data. These methods work with deformed and non-symmetric varied shaped objects and do not necessitate multiple objects for handling the missing wedge problem. The developed methods and contributions include: (a) an improved method for manual image segmentation, (b) a new approach to 'complete' the segmented and reconstructed incomplete 3D images, (c) a polyhedral structural distance model to predict the polyhedral shapes of these microstructures, (d) a new shape descriptor for polyhedral shapes, named as polyhedron profile statistic, and (e) the Bayes classifier, linear discriminant analysis and support vector machine based classifiers for supervised incomplete polyhedral shape classification. Finally, the predicted 3D shapes for these bacterial microstructures belong to the Johnson solids family, and these shapes along with their other geometric properties are important for better understanding of their chemical and biological characteristics.

Using interaction data for improving the offline and online evaluation of search engines

This thesis investigates how web search evaluation can be improved using historical interaction data. Modern search engines combine offline and online evaluation approaches in a sequence of steps that a tested change needs to pass through to be accepted as an improvement and subsequently deployed. We refer to such a sequence of steps as an evaluation pipeline. In this thesis, we consider the evaluation pipeline to contain three sequential steps: an offline evaluation step, an online evaluation scheduling step, and an online evaluation step. In this thesis we show that historical user interaction data can aid in improving the accuracy or efficiency of each of the steps of the web search evaluation pipeline. As a result of these improvements, the overall efficiency of the entire evaluation pipeline is increased. Firstly, we investigate how user interaction data can be used to build accurate offline evaluation methods for query auto-completion mechanisms. We propose a family of offline evaluation metrics for query auto-completion that represents the effort the user has to spend in order to submit their query. The parameters of our proposed metrics are trained against a set of user interactions recorded in the search engine’s query logs. From our experimental study, we observe that our proposed metrics are significantly more correlated with an online user satisfaction indicator than the metrics proposed in the existing literature. Hence, fewer changes will pass the offline evaluation step to be rejected after the online evaluation step. As a result, this would allow us to achieve a higher efficiency of the entire evaluation pipeline. Secondly, we state the problem of the optimised scheduling of online experiments. We tackle this problem by considering a greedy scheduler that prioritises the evaluation queue according to the predicted likelihood of success of a particular experiment. This predictor is trained on a set of online experiments, and uses a diverse set of features to represent an online experiment. Our study demonstrates that a higher number of successful experiments per unit of time can be achieved by deploying such a scheduler on the second step of the evaluation pipeline. Consequently, we argue that the efficiency of the evaluation pipeline can be increased. Next, to improve the efficiency of the online evaluation step, we propose the Generalised Team Draft interleaving framework. Generalised Team Draft considers both the interleaving policy (how often a particular combination of results is shown) and click scoring (how important each click is) as parameters in a data-driven optimisation of the interleaving sensitivity. Further, Generalised Team Draft is applicable beyond domains with a list-based representation of results, i.e. in domains with a grid-based representation, such as image search. Our study using datasets of interleaving experiments performed both in document and image search domains demonstrates that Generalised Team Draft achieves the highest sensitivity. A higher sensitivity indicates that the interleaving experiments can be deployed for a shorter period of time or use a smaller sample of users. Importantly, Generalised Team Draft optimises the interleaving parameters w.r.t. historical interaction data recorded in the interleaving experiments. Finally, we propose to apply the sequential testing methods to reduce the mean deployment time for the interleaving experiments. We adapt two sequential tests for the interleaving experimentation. We demonstrate that one can achieve a significant decrease in experiment duration by using such sequential testing methods. The highest efficiency is achieved by the sequential tests that adjust their stopping thresholds using historical interaction data recorded in diagnostic experiments. Our further experimental study demonstrates that cumulative gains in the online experimentation efficiency can be achieved by combining the interleaving sensitivity optimisation approaches, including Generalised Team Draft, and the sequential testing approaches. Overall, the central contributions of this thesis are the proposed approaches to improve the accuracy or efficiency of the steps of the evaluation pipeline: the offline evaluation frameworks for the query auto-completion, an approach for the optimised scheduling of online experiments, a general framework for the efficient online interleaving evaluation, and a sequential testing approach for the online search evaluation. The experiments in this thesis are based on massive real-life datasets obtained from Yandex, a leading commercial search engine. These experiments demonstrate the potential of the proposed approaches to improve the efficiency of the evaluation pipeline.

Desenvolvimento de ferramentas de análise de desempenho de atletas em 3 diferentes modalidades : Handebol, Provas Combinadas e Salto com Vara

Universidade Estadual de Campinas. Faculdade de Educação Física

Analise cinematica das trajetorias de jogadores de handebol obtidas por rastreamento automatico

Universidade Estadual de Campinas . Faculdade de Educação Física

A study of the validity and reliability of the Brazilian version of the Body Shape Questionnaire (BSQ) among adolescents

OBJECTIVES: to produce evidence of the validity and reliability of the Body Shape Questionnaire (BSQ) - a tool for measuring an individual's attitude towards his or her body image. METHODS: the study covered 386 young people of both sexes aged between 10 and 18 from a private school and used self-applied questionnaires and anthropometric evaluation. It evaluated the internal consistency, the discriminant validity for differences from the means, according to nutritional status (underweight, eutrophic, overweight and obese), the concurrent validity by way of Spearman's correlation coefficient between the scale and the Body Mass Index (BMI), the waist-hip circumference ratio (WHR) and the waist circumference (WC). Reliability was tested using Wilcoxon's Test, the intraclass correlation coefficient and the Bland-Altman figures. RESULTS: the BSQ displayed good internal consistency (±=0.96) and was capable of discriminating among the total population, boys and girls, according to nutritional status (p<0.001). It correlated with the BMI (r=0.41; p<0.001), WHR (r=-0.10; p=0.043) and WC (r=0.24; p<0.001) and its reliability was confirmed by intraclass correlation (r=0.91; p<0.001) for the total population. The questionnaire was easy to understand and could be completed quickly. CONCLUSIONS: the BSQ presented good results, thereby providing evidence of its validity and reliability. It is therefore recommended for evaluation of body image attitudes among adolescents.