Supervised classification of minimally invasive surgery psychomotor competence based on instruments motion analysis

Background Objective assessment of psychomotor skills has become an important challenge in the training of minimally invasive surgical (MIS) techniques. Currently, no gold standard defining surgical competence exists for classifying residents according to their surgical skills. Supervised classification has been proposed as a means for objectively establishing competence thresholds in psychomotor skills evaluation. This report presents a study comparing three classification methods for establishing their validity in a set of tasks for basic skills’ assessment. Methods Linear discriminant analysis (LDA), support vector machines (SVM), and adaptive neuro-fuzzy inference systems (ANFIS) were used. A total of 42 participants, divided into an experienced group (4 expert surgeons and 14 residents with >10 laparoscopic surgeries performed) and a nonexperienced group (16 students and 8 residents with <10 laparoscopic surgeries performed), performed three box trainer tasks validated for assessment of MIS psychomotor skills. Instrument movements were captured using the TrEndo tracking system, and nine motion analysis parameters (MAPs) were analyzed. The performance of the classifiers was measured by leave-one-out cross-validation using the scores obtained by the participants. Results The mean accuracy performances of the classifiers were 71 % (LDA), 78.2 % (SVM), and 71.7 % (ANFIS). No statistically significant differences in the performance were identified between the classifiers. Conclusions The three proposed classifiers showed good performance in the discrimination of skills, especially when information from all MAPs and tasks combined were considered. A correlation between the surgeons’ previous experience and their execution of the tasks could be ascertained from results. However, misclassifications across all the classifiers could imply the existence of other factors influencing psychomotor competence.

Determinación de un modelo de geoide gravimétrico para Puerto Rico como sistema de referencia para las altitudes ortométricas

El geoide, definido como la superficie equipotencial que mejor se ajusta (en el sentido de los mínimos cuadrados) al nivel medio del mar en una determinada época, es la superficie que utilizamos como referencia para determinar las altitudes ortométricas. Si disponemos de una superficie equipotencial de referencia como dátum altimétrico preciso o geoide local, podemos entonces determinar las altitudes ortométricas de forma eficiente a partir de las altitudes elipsoidales proporcionadas por el Sistema Global de Navegación por Satélite (Global Navigation Satellite System, GNSS ). Como es sabido uno de los problemas no resueltos de la geodesia (quizás el más importante de los mismos en la actualidad) es la carencia de un dátum altimétrico global (Sjoberg, 2011) con las precisiones adecuadas. Al no existir un dátum altimétrico global que nos permita obtener los valores absolutos de la ondulación del geoide con la precisión requerida, es necesario emplear modelos geopotenciales como alternativa. Recientemente fue publicado el modelo EGM2008 en el que ha habido una notable mejoría de sus tres fuentes de datos, por lo que este modelo contiene coeficientes adicionales hasta el grado 2190 y orden 2159 y supone una sustancial mejora en la precisión (Pavlis et al., 2008). Cuando en una región determinada se dispone de valores de gravedad y Modelos Digitales del Terreno (MDT) de calidad, es posible obtener modelos de superficies geopotenciales más precisos y de mayor resolución que los modelos globales. Si bien es cierto que el Servicio Nacional Geodésico de los Estados Unidos de América (National Geodetic Survey, NGS) ha estado desarrollando modelos del geoide para la región de los Estados Unidos de América continentales y todos sus territorios desde la década de los noventa, también es cierto que las zonas de Puerto Rico y las Islas Vírgenes Estadounidenses han quedado un poco rezagadas al momento de poder aplicar y obtener resultados de mayor precisión con estos modelos regionales del geoide. En la actualidad, el modelo geopotencial regional vigente para la zona de Puerto Rico y las Islas Vírgenes Estadounidenses es el GEOID12A (Roman y Weston, 2012). Dada la necesidad y ante la incertidumbre de saber cuál sería el comportamiento de un modelo del geoide desarrollado única y exclusivamente con datos de gravedad locales, nos hemos dado a la tarea de desarrollar un modelo de geoide gravimétrico como sistema de referencia para las altitudes ortométricas. Para desarrollar un modelo del geoide gravimétrico en la isla de Puerto Rico, fue necesario implementar una metodología que nos permitiera analizar y validar los datos de gravedad terrestre existentes. Utilizando validación por altimetría con sistemas de información geográfica y validación matemática por colocación con el programa Gravsoft (Tscherning et al., 1994) en su modalidad en Python (Nielsen et al., 2012), fue posible validar 1673 datos de anomalías aire libre de un total de 1894 observaciones obtenidas de la base de datos del Bureau Gravimétrico Internacional (BGI). El aplicar estas metodologías nos permitió obtener una base de datos anomalías de la gravedad fiable la cual puede ser utilizada para una gran cantidad de aplicaciones en ciencia e ingeniería. Ante la poca densidad de datos de gravedad existentes, fue necesario emplear un método alternativo para densificar los valores de anomalías aire libre existentes. Empleando una metodología propuesta por Jekeli et al. (2009b) se procedió a determinar anomalías aire libre a partir de los datos de un MDT. Estas anomalías fueron ajustadas utilizando las anomalías aire libre validadas y tras aplicar un ajuste de mínimos cuadrados por zonas geográficas, fue posible obtener una malla de datos de anomalías aire libre uniforme a partir de un MDT. Tras realizar las correcciones topográficas, determinar el efecto indirecto de la topografía del terreno y la contribución del modelo geopotencial EGM2008, se obtuvo una malla de anomalías residuales. Estas anomalías residuales fueron utilizadas para determinar el geoide gravimétrico utilizando varias técnicas entre las que se encuentran la aproximación plana de la función de Stokes y las modificaciones al núcleo de Stokes, propuestas por Wong y Gore (1969), Vanicek y Kleusberg (1987) y Featherstone et al. (1998). Ya determinados los distintos modelos del geoide gravimétrico, fue necesario validar los mismos y para eso se utilizaron una serie de estaciones permanentes de la red de nivelación del Datum Vertical de Puerto Rico de 2002 (Puerto Rico Vertical Datum 2002, PRVD02 ), las cuales tenían publicados sus valores de altitud elipsoidal y elevación. Ante la ausencia de altitudes ortométricas en las estaciones permanentes de la red de nivelación, se utilizaron las elevaciones obtenidas a partir de nivelación de primer orden para determinar los valores de la ondulación del geoide geométrico (Roman et al., 2013). Tras establecer un total de 990 líneas base, se realizaron dos análisis para determinar la 'precisión' de los modelos del geoide. En el primer análisis, que consistió en analizar las diferencias entre los incrementos de la ondulación del geoide geométrico y los incrementos de la ondulación del geoide de los distintos modelos (modelos gravimétricos, EGM2008 y GEOID12A) en función de las distancias entre las estaciones de validación, se encontró que el modelo con la modificación del núcleo de Stokes propuesta por Wong y Gore presentó la mejor 'precisión' en un 91,1% de los tramos analizados. En un segundo análisis, en el que se consideraron las 990 líneas base, se determinaron las diferencias entre los incrementos de la ondulación del geoide geométrico y los incrementos de la ondulación del geoide de los distintos modelos (modelos gravimétricos, EGM2008 y GEOID12A), encontrando que el modelo que presenta la mayor 'precisión' también era el geoide con la modificación del núcleo de Stokes propuesta por Wong y Gore. En este análisis, el modelo del geoide gravimétrico de Wong y Gore presento una 'precisión' de 0,027 metros en comparación con la 'precisión' del modelo EGM2008 que fue de 0,031 metros mientras que la 'precisión' del modelo regional GEOID12A fue de 0,057 metros. Finalmente podemos decir que la metodología aquí presentada es una adecuada ya que fue posible obtener un modelo del geoide gravimétrico que presenta una mayor 'precisión' que los modelos geopotenciales disponibles, incluso superando la precisión del modelo geopotencial global EGM2008. ABSTRACT The geoid, defined as the equipotential surface that best fits (in the least squares sense) to the mean sea level at a particular time, is the surface used as a reference to determine the orthometric heights. If we have an equipotential reference surface or a precise local geoid, we can then determine the orthometric heights efficiently from the ellipsoidal heights, provided by the Global Navigation Satellite System (GNSS). One of the most common and important an unsolved problem in geodesy is the lack of a global altimetric datum (Sjoberg, 2011)) with the appropriate precision. In the absence of one which allows us to obtain the absolute values of the geoid undulation with the required precision, it is necessary to use alternative geopotential models. The EGM2008 was recently published, in which there has been a marked improvement of its three data sources, so this model contains additional coefficients of degree up to 2190 and order 2159, and there is a substantial improvement in accuracy (Pavlis et al., 2008). When a given region has gravity values and high quality digital terrain models (DTM), it is possible to obtain more accurate regional geopotential models, with a higher resolution and precision, than global geopotential models. It is true that the National Geodetic Survey of the United States of America (NGS) has been developing geoid models for the region of the continental United States of America and its territories from the nineties, but which is also true is that areas such as Puerto Rico and the U.S. Virgin Islands have lagged behind when to apply and get more accurate results with these regional geopotential models. Right now, the available geopotential model for Puerto Rico and the U.S. Virgin Islands is the GEOID12A (Roman y Weston, 2012). Given this need and given the uncertainty of knowing the behavior of a regional geoid model developed exclusively with data from local gravity, we have taken on the task of developing a gravimetric geoid model to use as a reference system for orthometric heights. To develop a gravimetric geoid model in the island of Puerto Rico, implementing a methodology that allows us to analyze and validate the existing terrestrial gravity data is a must. Using altimetry validation with GIS and mathematical validation by collocation with the Gravsoft suite programs (Tscherning et al., 1994) in its Python version (Nielsen et al., 2012), it was possible to validate 1673 observations with gravity anomalies values out of a total of 1894 observations obtained from the International Bureau Gravimetric (BGI ) database. Applying these methodologies allowed us to obtain a database of reliable gravity anomalies, which can be used for many applications in science and engineering. Given the low density of existing gravity data, it was necessary to employ an alternative method for densifying the existing gravity anomalies set. Employing the methodology proposed by Jekeli et al. (2009b) we proceeded to determine gravity anomaly data from a DTM. These anomalies were adjusted by using the validated free-air gravity anomalies and, after that, applying the best fit in the least-square sense by geographical area, it was possible to obtain a uniform grid of free-air anomalies obtained from a DTM. After applying the topographic corrections, determining the indirect effect of topography and the contribution of the global geopotential model EGM2008, a grid of residual anomalies was obtained. These residual anomalies were used to determine the gravimetric geoid by using various techniques, among which are the planar approximation of the Stokes function and the modifications of the Stokes kernel, proposed by Wong y Gore (1969), Vanicek y Kleusberg (1987) and Featherstone et al. (1998). After determining the different gravimetric geoid models, it was necessary to validate them by using a series of stations of the Puerto Rico Vertical Datum of 2002 (PRVD02) leveling network. These stations had published its values of ellipsoidal height and elevation, and in the absence of orthometric heights, we use the elevations obtained from first - order leveling to determine the geometric geoid undulation (Roman et al., 2013). After determine a total of 990 baselines, two analyzes were performed to determine the ' accuracy ' of the geoid models. The first analysis was to analyze the differences between the increments of the geometric geoid undulation with the increments of the geoid undulation of the different geoid models (gravimetric models, EGM2008 and GEOID12A) in function of the distance between the validation stations. Through this analysis, it was determined that the model with the modified Stokes kernel given by Wong and Gore had the best 'accuracy' in 91,1% for the analyzed baselines. In the second analysis, in which we considered the 990 baselines, we analyze the differences between the increments of the geometric geoid undulation with the increments of the geoid undulation of the different geoid models (gravimetric models, EGM2008 and GEOID12A) finding that the model with the highest 'accuracy' was also the model with modifying Stokes kernel given by Wong and Gore. In this analysis, the Wong and Gore gravimetric geoid model presented an 'accuracy' of 0,027 meters in comparison with the 'accuracy' of global geopotential model EGM2008, which gave us an 'accuracy' of 0,031 meters, while the 'accuracy ' of the GEOID12A regional model was 0,057 meters. Finally we can say that the methodology presented here is adequate as it was possible to obtain a gravimetric geoid model that has a greater 'accuracy' than the geopotential models available, even surpassing the accuracy of global geopotential model EGM2008.

Development of a methodology for nuclear data uncertainty propagation on isotopic evolution calculations for advanced nuclear systems

Una apropiada evaluación de los márgenes de seguridad de una instalación nuclear, por ejemplo, una central nuclear, tiene en cuenta todas las incertidumbres que afectan a los cálculos de diseño, funcionanmiento y respuesta ante accidentes de dicha instalación. Una fuente de incertidumbre son los datos nucleares, que afectan a los cálculos neutrónicos, de quemado de combustible o activación de materiales. Estos cálculos permiten la evaluación de las funciones respuesta esenciales para el funcionamiento correcto durante operación, y también durante accidente. Ejemplos de esas respuestas son el factor de multiplicación neutrónica o el calor residual después del disparo del reactor. Por tanto, es necesario evaluar el impacto de dichas incertidumbres en estos cálculos. Para poder realizar los cálculos de propagación de incertidumbres, es necesario implementar metodologías que sean capaces de evaluar el impacto de las incertidumbres de estos datos nucleares. Pero también es necesario conocer los datos de incertidumbres disponibles para ser capaces de manejarlos. Actualmente, se están invirtiendo grandes esfuerzos en mejorar la capacidad de analizar, manejar y producir datos de incertidumbres, en especial para isótopos importantes en reactores avanzados. A su vez, nuevos programas/códigos están siendo desarrollados e implementados para poder usar dichos datos y analizar su impacto. Todos estos puntos son parte de los objetivos del proyecto europeo ANDES, el cual ha dado el marco de trabajo para el desarrollo de esta tesis doctoral. Por tanto, primero se ha llevado a cabo una revisión del estado del arte de los datos nucleares y sus incertidumbres, centrándose en los tres tipos de datos: de decaimiento, de rendimientos de fisión y de secciones eficaces. A su vez, se ha realizado una revisión del estado del arte de las metodologías para la propagación de incertidumbre de estos datos nucleares. Dentro del Departamento de Ingeniería Nuclear (DIN) se propuso una metodología para la propagación de incertidumbres en cálculos de evolución isotópica, el Método Híbrido. Esta metodología se ha tomado como punto de partida para esta tesis, implementando y desarrollando dicha metodología, así como extendiendo sus capacidades. Se han analizado sus ventajas, inconvenientes y limitaciones. El Método Híbrido se utiliza en conjunto con el código de evolución isotópica ACAB, y se basa en el muestreo por Monte Carlo de los datos nucleares con incertidumbre. En esta metodología, se presentan diferentes aproximaciones según la estructura de grupos de energía de las secciones eficaces: en un grupo, en un grupo con muestreo correlacionado y en multigrupos. Se han desarrollado diferentes secuencias para usar distintas librerías de datos nucleares almacenadas en diferentes formatos: ENDF-6 (para las librerías evaluadas), COVERX (para las librerías en multigrupos de SCALE) y EAF (para las librerías de activación). Gracias a la revisión del estado del arte de los datos nucleares de los rendimientos de fisión se ha identificado la falta de una información sobre sus incertidumbres, en concreto, de matrices de covarianza completas. Además, visto el renovado interés por parte de la comunidad internacional, a través del grupo de trabajo internacional de cooperación para evaluación de datos nucleares (WPEC) dedicado a la evaluación de las necesidades de mejora de datos nucleares mediante el subgrupo 37 (SG37), se ha llevado a cabo una revisión de las metodologías para generar datos de covarianza. Se ha seleccionando la actualización Bayesiana/GLS para su implementación, y de esta forma, dar una respuesta a dicha falta de matrices completas para rendimientos de fisión. Una vez que el Método Híbrido ha sido implementado, desarrollado y extendido, junto con la capacidad de generar matrices de covarianza completas para los rendimientos de fisión, se han estudiado diferentes aplicaciones nucleares. Primero, se estudia el calor residual tras un pulso de fisión, debido a su importancia para cualquier evento después de la parada/disparo del reactor. Además, se trata de un ejercicio claro para ver la importancia de las incertidumbres de datos de decaimiento y de rendimientos de fisión junto con las nuevas matrices completas de covarianza. Se han estudiado dos ciclos de combustible de reactores avanzados: el de la instalación europea para transmutación industrial (EFIT) y el del reactor rápido de sodio europeo (ESFR), en los cuales se han analizado el impacto de las incertidumbres de los datos nucleares en la composición isotópica, calor residual y radiotoxicidad. Se han utilizado diferentes librerías de datos nucleares en los estudios antreriores, comparando de esta forma el impacto de sus incertidumbres. A su vez, mediante dichos estudios, se han comparando las distintas aproximaciones del Método Híbrido y otras metodologías para la porpagación de incertidumbres de datos nucleares: Total Monte Carlo (TMC), desarrollada en NRG por A.J. Koning y D. Rochman, y NUDUNA, desarrollada en AREVA GmbH por O. Buss y A. Hoefer. Estas comparaciones demostrarán las ventajas del Método Híbrido, además de revelar sus limitaciones y su rango de aplicación. ABSTRACT For an adequate assessment of safety margins of nuclear facilities, e.g. nuclear power plants, it is necessary to consider all possible uncertainties that affect their design, performance and possible accidents. Nuclear data are a source of uncertainty that are involved in neutronics, fuel depletion and activation calculations. These calculations can predict critical response functions during operation and in the event of accident, such as decay heat and neutron multiplication factor. Thus, the impact of nuclear data uncertainties on these response functions needs to be addressed for a proper evaluation of the safety margins. Methodologies for performing uncertainty propagation calculations need to be implemented in order to analyse the impact of nuclear data uncertainties. Nevertheless, it is necessary to understand the current status of nuclear data and their uncertainties, in order to be able to handle this type of data. Great eórts are underway to enhance the European capability to analyse/process/produce covariance data, especially for isotopes which are of importance for advanced reactors. At the same time, new methodologies/codes are being developed and implemented for using and evaluating the impact of uncertainty data. These were the objectives of the European ANDES (Accurate Nuclear Data for nuclear Energy Sustainability) project, which provided a framework for the development of this PhD Thesis. Accordingly, first a review of the state-of-the-art of nuclear data and their uncertainties is conducted, focusing on the three kinds of data: decay, fission yields and cross sections. A review of the current methodologies for propagating nuclear data uncertainties is also performed. The Nuclear Engineering Department of UPM has proposed a methodology for propagating uncertainties in depletion calculations, the Hybrid Method, which has been taken as the starting point of this thesis. This methodology has been implemented, developed and extended, and its advantages, drawbacks and limitations have been analysed. It is used in conjunction with the ACAB depletion code, and is based on Monte Carlo sampling of variables with uncertainties. Different approaches are presented depending on cross section energy-structure: one-group, one-group with correlated sampling and multi-group. Differences and applicability criteria are presented. Sequences have been developed for using different nuclear data libraries in different storing-formats: ENDF-6 (for evaluated libraries) and COVERX (for multi-group libraries of SCALE), as well as EAF format (for activation libraries). A revision of the state-of-the-art of fission yield data shows inconsistencies in uncertainty data, specifically with regard to complete covariance matrices. Furthermore, the international community has expressed a renewed interest in the issue through the Working Party on International Nuclear Data Evaluation Co-operation (WPEC) with the Subgroup (SG37), which is dedicated to assessing the need to have complete nuclear data. This gives rise to this review of the state-of-the-art of methodologies for generating covariance data for fission yields. Bayesian/generalised least square (GLS) updating sequence has been selected and implemented to answer to this need. Once the Hybrid Method has been implemented, developed and extended, along with fission yield covariance generation capability, different applications are studied. The Fission Pulse Decay Heat problem is tackled first because of its importance during events after shutdown and because it is a clean exercise for showing the impact and importance of decay and fission yield data uncertainties in conjunction with the new covariance data. Two fuel cycles of advanced reactors are studied: the European Facility for Industrial Transmutation (EFIT) and the European Sodium Fast Reactor (ESFR), and response function uncertainties such as isotopic composition, decay heat and radiotoxicity are addressed. Different nuclear data libraries are used and compared. These applications serve as frameworks for comparing the different approaches of the Hybrid Method, and also for comparing with other methodologies: Total Monte Carlo (TMC), developed at NRG by A.J. Koning and D. Rochman, and NUDUNA, developed at AREVA GmbH by O. Buss and A. Hoefer. These comparisons reveal the advantages, limitations and the range of application of the Hybrid Method.

Towards downscaling of aerosol gridded dataset for improving solar resource assessment, an application to Spain

Solar radiation estimates with clear sky models require estimations of aerosol data. The low spatial resolution of current aerosol datasets, with their remarkable drift from measured data, poses a problem in solar resource estimation. This paper proposes a new downscaling methodology by combining support vector machines for regression (SVR) and kriging with external drift, with data from the MACC reanalysis datasets and temperature and rainfall measurements from 213 meteorological stations in continental Spain. The SVR technique was proven efficient in aerosol variable modeling. The Linke turbidity factor (TL) and the aerosol optical depth at 550 nm (AOD 550) estimated with SVR generated significantly lower errors in AERONET positions than MACC reanalysis estimates. The TL was estimated with relative mean absolute error (rMAE) of 10.2% (compared with AERONET), against the MACC rMAE of 18.5%. A similar behavior was seen with AOD 550, estimated with rMAE of 8.6% (compared with AERONET), against the MACC rMAE of 65.6%. Kriging using MACC data as an external drift was found useful in generating high resolution maps (0.05° × 0.05°) of both aerosol variables. We created high resolution maps of aerosol variables in continental Spain for the year 2008. The proposed methodology was proven to be a valuable tool to create high resolution maps of aerosol variables (TL and AOD 550). This methodology shows meaningful improvements when compared with estimated available databases and therefore, leads to more accurate solar resource estimations. This methodology could also be applied to the prediction of other atmospheric variables, whose datasets are of low resolution.

High-quality region-based foreground segmentation using a spatial grid of SVM classifiers

This paper presents a novel background modeling system that uses a spatial grid of Support Vector Machines classifiers for segmenting moving objects, which is a key step in many video-based consumer applications. The system is able to adapt to a large range of dynamic background situations since no parametric model or statistical distribution are assumed. This is achieved by using a different classifier per image region that learns the specific appearance of that scene region and its variations (illumination changes, dynamic backgrounds, etc.). The proposed system has been tested with a recent public database, outperforming other state-of-the-art algorithms.

Sistemas de ayuda al diagnóstico y reconocimiento de microcalcificaciones en mamografía mediante descriptores de escala y redes jerárquicas

La presente Tesis investiga el campo del reconocimiento automático de imágenes mediante ordenador aplicado al análisis de imágenes médicas en mamografía digital. Hay un interés por desarrollar sistemas de aprendizaje que asistan a los radiólogos en el reconocimiento de las microcalcificaciones para apoyarles en los programas de cribado y prevención del cáncer de mama. Para ello el análisis de las microcalcificaciones se ha revelado como técnica clave de diagnóstico precoz, pero sin embargo el diseño de sistemas automáticos para reconocerlas es complejo por la variabilidad y condiciones de las imágenes mamográficas. En este trabajo se analizan los planteamientos teóricos de diseño de sistemas de reconocimiento de imágenes, con énfasis en los problemas específicos de detección y clasificación de microcalcificaciones. Se ha realizado un estudio que incluye desde las técnicas de operadores morfológicos, redes neuronales, máquinas de vectores soporte, hasta las más recientes de aprendizaje profundo mediante redes neuronales convolucionales, contemplando la importancia de los conceptos de escala y jerarquía a la hora del diseño y sus implicaciones en la búsqueda de la arquitectura de conexiones y capas de la red. Con estos fundamentos teóricos y elementos de diseño procedentes de otros trabajos en este área realizados por el autor, se implementan tres sistemas de reconocimiento de mamografías que reflejan una evolución tecnológica, culminando en un sistema basado en Redes Neuronales Convolucionales (CNN) cuya arquitectura se diseña gracias al análisis teórico anterior y a los resultados prácticos de análisis de escalas llevados a cabo en nuestra base de datos de imágenes. Los tres sistemas se entrenan y validan con la base de datos de mamografías DDSM, con un total de 100 muestras de entrenamiento y 100 de prueba escogidas para evitar sesgos y reflejar fielmente un programa de cribado. La validez de las CNN para el problema que nos ocupa queda demostrada y se propone un camino de investigación para el diseño de su arquitectura. ABSTRACT This Dissertation investigates the field of computer image recognition applied to medical imaging in mammography. There is an interest in developing learning systems to assist radiologists in recognition of microcalcifications to help them in screening programs for prevention of breast cancer. Analysis of microcalcifications has emerged as a key technique for early diagnosis of breast cancer, but the design of automatic systems to recognize them is complicated by the variability and conditions of mammographic images. In this Thesis the theoretical approaches to design image recognition systems are discussed, with emphasis on the specific problems of detection and classification of microcalcifications. Our study includes techniques ranging from morphological operators, neural networks and support vector machines, to the most recent deep convolutional neural networks. We deal with learning theory by analyzing the importance of the concepts of scale and hierarchy at the design stage and its implications in the search for the architecture of connections and network layers. With these theoretical facts and design elements coming from other works in this area done by the author, three mammogram recognition systems which reflect technological developments are implemented, culminating in a system based on Convolutional Neural Networks (CNN), whose architecture is designed thanks to the previously mentioned theoretical study and practical results of analysis conducted on scales in our image database. All three systems are trained and validated against the DDSM mammographic database, with a total of 100 training samples and 100 test samples chosen to avoid bias and stand for a real screening program. The validity of the CNN approach to the problem is demonstrated and a research way to help in designing the architecture of these networks is proposed.

Determination of chromite composition from multispectral reflectance measurements (400-1000 nm)

The relationship between the chemical composition and the multispectral reflectance values of chromite in the VNIR (Visible and Near-Infra-Red) realm is tested and mathematically analysed. Statisticaltools as Pearson's correlation coefficients, linear stepwise regression analysis and least-square adjustments are applied to two populations of data obtained from 14 selected samples 01 chromite multielemental microprobe analysis and multispectral reflectance values (400-1 000 nm). Results show that both data sets correlate, and suggest that the VNIR reflectance spectra can be used as a tool to determine the chemical composition of chromites.

GESTÃO DO CONHECIMENTO: UM ESTUDO SOBRE A RELAÇÃO ENTRE A LIDERANÇA TRANSFORMACIONAL, A CONVERSÃO DO CONHECIMENTO E A EFICÁCIA ORGANIZACIONAL

Este estudo teve como objetivo principal analisar a relação entre a Liderança Transformacional, a Conversão do Conhecimento e a Eficácia Organizacional. Foram considerados como pressupostos teóricos conceitos consolidados sobre os temas desta relação, além de recentes pesquisas já realizadas em outros países e contextos organizacionais. Com base nisto identificou-se potencial estudo de um modelo que relacionasse estes três conceitos. Para tal considera-se que as organizações que buscam atingir Vantagem Competitiva e incorporam a Knowledge-Based View possam conquistar diferenciação frente a seus concorrentes. Nesse contexto o conhecimento ganha maior destaque e papel protagonista nestas organizações. Dessa forma criar conhecimento através de seus colaboradores, passa a ser um dos desafios dessas organizações ao passo que sugere melhoria de seus indicadores Econômicos, Sociais, Sistêmicos e Políticos, o que se define por Eficácia Organizacional. Portanto os modos de conversão do conhecimento nas organizações, demonstram relevância, uma vez que se cria e se converte conhecimentos através da interação entre o conhecimento existente de seus colaboradores. Essa conversão do conhecimento ou modelo SECI possui quatro modos que são a Socialização, Externalização, Combinação e Internalização. Nessa perspectiva a liderança nas organizações apresenta-se como um elemento capaz de influenciar seus colaboradores, propiciando maior dinâmica ao modelo SECI de conversão do conhecimento. Se identifica então na liderança do tipo Transformacional, características que possam influenciar colaboradores e entende-se que esta relação entre a Liderança Transformacional e a Conversão do Conhecimento possa ter influência positiva nos indicadores da Eficácia Organizacional. Dessa forma esta pesquisa buscou analisar um modelo que explorasse essa relação entre a liderança do tipo Transformacional, a Conversão do Conhecimento (SECI) e a Eficácia Organizacional. Esta pesquisa teve o caráter quantitativo com coleta de dados através do método survey, obtendo um total de 230 respondentes válidos de diferentes organizações. O instrumento de coleta de dados foi composto por afirmativas relativas ao modelo de relação pesquisado com um total de 44 itens. O perfil de respondentes concentrou-se entre 30 e 39 anos de idade, com a predominância de organizações privadas e de departamentos de TI/Telecom, Docência e Recursos Humanos respectivamente. O tratamento dos dados foi através da Análise Fatorial Exploratória e Modelagem de Equações Estruturais via Partial Least Square Path Modeling (PLS-PM). Como resultado da análise desta pesquisa, as hipóteses puderam ser confirmadas, concluindo que a Liderança Transformacional apresenta influência positiva nos modos de Conversão do Conhecimento e que; a Conversão do Conhecimento influencia positivamente na Eficácia Organizacional. Ainda, concluiu-se que a percepção entre os respondentes não apresenta resultado diferente sobre o modelo desta pesquisa entre quem possui ou não função de liderança.

Otimização termodinâmica do sistema binário Ti-Si.

A otimização de sistemas do tipo Ti-Si-X requer que os sistemas binários estejam constantemente atualizados. O sistema Ti-Si foi investigado experimentalmente desde a década de 50 e poucos estudos usaram os dados experimentais para calcular o diagrama de fases Ti-Si usando modelamento termodinâmico. A otimização mais recente do sistema Ti-Si foi realizada em 1998, descrevendo a fase Ti5Si3 como um intermetálico não estequiométrico contendo três sub-redes e mostrando a presença da fase intermetálica estequiométrica Ti3Si. Dada a recente disputa sobre a cinética de precipitação e a estabilidade das fases Ti3Si e Ti5Si3 nos sistemas Ti-Si e Ti-Si-X, o canto rico em titânio do sistema Ti-Si (estável e metaestável) foi otimizado no presente trabalho. Os limites de estabilidade de fases, os valores dos erros pelo método dos mínimos quadrados do procedimento de otimização e os desvios padrões relativos das variáveis calculadas foram discutidos para inspirar a realização de mais trabalhos experimentais para investigar as reações eutetóides estáveis e/ou metaestáveis, ?->? + Ti3Si e ?->? + + Ti5Si3; e para melhorar cada vez mais as otimizações termodinâmicas do diagrama de fases do sistema Ti-Si.

Real time optimization in chemical process: evaluation of strategies, improvements and industrial application.

The increasing economic competition drives the industry to implement tools that improve their processes efficiencies. The process automation is one of these tools, and the Real Time Optimization (RTO) is an automation methodology that considers economic aspects to update the process control in accordance with market prices and disturbances. Basically, RTO uses a steady-state phenomenological model to predict the process behavior, and then, optimizes an economic objective function subject to this model. Although largely implemented in industry, there is not a general agreement about the benefits of implementing RTO due to some limitations discussed in the present work: structural plant/model mismatch, identifiability issues and low frequency of set points update. Some alternative RTO approaches have been proposed in literature to handle the problem of structural plant/model mismatch. However, there is not a sensible comparison evaluating the scope and limitations of these RTO approaches under different aspects. For this reason, the classical two-step method is compared to more recently derivative-based methods (Modifier Adaptation, Integrated System Optimization and Parameter estimation, and Sufficient Conditions of Feasibility and Optimality) using a Monte Carlo methodology. The results of this comparison show that the classical RTO method is consistent, providing a model flexible enough to represent the process topology, a parameter estimation method appropriate to handle measurement noise characteristics and a method to improve the sample information quality. At each iteration, the RTO methodology updates some key parameter of the model, where it is possible to observe identifiability issues caused by lack of measurements and measurement noise, resulting in bad prediction ability. Therefore, four different parameter estimation approaches (Rotational Discrimination, Automatic Selection and Parameter estimation, Reparametrization via Differential Geometry and classical nonlinear Least Square) are evaluated with respect to their prediction accuracy, robustness and speed. The results show that the Rotational Discrimination method is the most suitable to be implemented in a RTO framework, since it requires less a priori information, it is simple to be implemented and avoid the overfitting caused by the Least Square method. The third RTO drawback discussed in the present thesis is the low frequency of set points update, this problem increases the period in which the process operates at suboptimum conditions. An alternative to handle this problem is proposed in this thesis, by integrating the classic RTO and Self-Optimizing control (SOC) using a new Model Predictive Control strategy. The new approach demonstrates that it is possible to reduce the problem of low frequency of set points updates, improving the economic performance. Finally, the practical aspects of the RTO implementation are carried out in an industrial case study, a Vapor Recompression Distillation (VRD) process located in Paulínea refinery from Petrobras. The conclusions of this study suggest that the model parameters are successfully estimated by the Rotational Discrimination method; the RTO is able to improve the process profit in about 3%, equivalent to 2 million dollars per year; and the integration of SOC and RTO may be an interesting control alternative for the VRD process.

Morphological classification of microtidal sand and gravel beaches

A new classification of microtidal sand and gravel beaches with very different morphologies is presented below. In 557 studied transects, 14 variables were used. Among the variables to be emphasized is the depth of the Posidonia oceanica. The classification was performed for 9 types of beaches: Type 1: Sand and gravel beaches, Type 2: Sand and gravel separated beaches, Type 3: Gravel and sand beaches, Type 4: Gravel and sand separated beaches, Type 5: Pure gravel beaches, Type 6: Open sand beaches, Type 7: Supported sand beaches, Type 8: Bisupported sand beaches and Type 9: Enclosed beaches. For the classification, several tools were used: discriminant analysis, neural networks and Support Vector Machines (SVM), the results were then compared. As there is no theory for deciding which is the most convenient neural network architecture to deal with a particular data set, an experimental study was performed with different numbers of neuron in the hidden layer. Finally, an architecture with 30 neurons was chosen. Different kernels were employed for SVM (Linear, Polynomial, Radial basis function and Sigmoid). The results obtained for the discriminant analysis were not as good as those obtained for the other two methods (ANN and SVM) which showed similar success.

Dissolved organic carbon in sediments of the central Arctic Ocean collected during POLARSTERN cruise ARK-XXVII/3 from August-September 2012

Marine organic matter (OM) sinks from surface waters to the seafloor via the biological pump. Benthic communities, which use this sedimented OM as energy and carbon source, produce dissolved organic matter (DOM) in the process of remineralization, enriching the sediment porewater with fresh DOM compounds. We hypothesized that in the oligotrophic deep Arctic basin the molecular signal of freshly deposited primary produced OM is restricted to the surface sediment pore waters which should differ from bottom water and deeper sediment pore water in DOM composition. This study focused on: 1) the molecular composition of the DOM in sediment pore waters of the deep Eurasian Arctic basins, 2) whether the signal of marine vs. terrigenous DOM is represented by different compounds preserved in the sediment pore waters and 3) whether there is any relation between Arctic Ocean ice cover and DOM composition. Molecular data, obtained via 15 Tesla Fourier transform ion cyclotron resonance mass spectrometer, were correlated with environmental parameters by partial least square analysis. The fresher marine detrital OM signal from surface waters was limited to pore waters from < 5 cm sediment depth. The productive ice margin stations showed higher abundances of peptides, unsaturated aliphatics and saturated fatty acids formulae, indicative of fresh OM/pigments deposition, compared to northernmost stations which had stronger aromatic signals. This study contributes to the understanding of the coupling between the Arctic Ocean productivity and its depositional regime, and how it will be altered in response to sea ice retreat and increasing river runoff.

Efficacy of sertraline in a 12-week trial for generalized anxiety disorder

Objective: Sertraline's efficacy and tolerability in treating generalized anxiety disorder were evaluated. Method: Adult outpatients with DSM-IV generalized anxiety disorder and a total score of 18 or higher on the Hamilton Anxiety Rating Scale were eligible. After a 1-week single-blind placebo lead-in, patients were randomly assigned to 12 weeks of double-blind treatment with placebo (N=188, mean baseline anxiety score=25) or flexible doses (50-150 mg/day) of sertraline (N=182, mean anxiety score=25). The primary outcome measure was baseline-to-endpoint change in the Hamilton anxiety scale total score. A secondary efficacy measure was the Clinical Global Impression (CGI) improvement score; response was defined as a score of 2 or less. Results: Sertraline patients had significantly greater improvement than placebo patients on all efficacy measures at week 4. Analysis of covariance of the intent-to-treat group at endpoint (with the last observation carried forward) showed a significant difference in the decrease from baseline of the least-square mean total score on the Hamilton anxiety scale between sertraline (mean=11.7) and placebo (mean=8.0). Significantly greater endpoint improvement with sertraline than placebo was obtained for mean scores on the Hamilton anxiety scale psychic factor (6.7 versus 4.1) and somatic factor (5.0 versus 3.9). The rate of responders, based on CGI improvement and last observation carried forward, was significantly higher for sertraline (63%) than placebo (37%). Sertraline was well tolerated; 8% of patients versus 10% for placebo dropped out because of adverse events. Conclusions: Sertraline appears to be efficacious and well tolerated in the treatment of generalized anxiety disorder.

Prediction of subcellular localization using sequence-biased recurrent networks

Motivation: Targeting peptides direct nascent proteins to their specific subcellular compartment. Knowledge of targeting signals enables informed drug design and reliable annotation of gene products. However, due to the low similarity of such sequences and the dynamical nature of the sorting process, the computational prediction of subcellular localization of proteins is challenging. Results: We contrast the use of feed forward models as employed by the popular TargetP/SignalP predictors with a sequence-biased recurrent network model. The models are evaluated in terms of performance at the residue level and at the sequence level, and demonstrate that recurrent networks improve the overall prediction performance. Compared to the original results reported for TargetP, an ensemble of the tested models increases the accuracy by 6 and 5% on non-plant and plant data, respectively.

Predicting the solvent accessibility of transmembrane residues from protein sequence

In this study, we propose a novel method to predict the solvent accessible surface areas of transmembrane residues. For both transmembrane alpha-helix and beta-barrel residues, the correlation coefficients between the predicted and observed accessible surface areas are around 0.65. On the basis of predicted accessible surface areas, residues exposed to the lipid environment or buried inside a protein can be identified by using certain cutoff thresholds. We have extensively examined our approach based on different definitions of accessible surface areas and a variety of sets of control parameters. Given that experimentally determining the structures of membrane proteins is very difficult and membrane proteins are actually abundant in nature, our approach is useful for theoretically modeling membrane protein tertiary structures, particularly for modeling the assembly of transmembrane domains. This approach can be used to annotate the membrane proteins in proteomes to provide extra structural and functional information.