Breast cancer continues to be the most frequently diagnosed cancer in Canadian women. Currently, mammography is the clinically accepted best modality for breast cancer detection and the regular use of screening has been shown to contribute to the reduced mortality. However, mammography suffers from several drawbacks which limit its sensitivity and specificity. As a potential solution, digital breast tomosynthesis (DBT) uses a limited number (typically 10-20) of low-dose x-ray projections to produce a three-dimensional tomographic representation of the breast. The reconstruction of DBT images is challenged by such incomplete sampling. The purpose of this thesis is to evaluate the effect of image acquisition parameters on image quality of DBT for various reconstruction techniques and to optimize these, with three specific goals: A) Develop a better power spectrum estimator for detectability calculation as a task-based image quality index; B) Develop a paired-view algorithm for artifact removal in DBT reconstruction; and C) Increase dose efficiency in DBT by reducing random noise. A better power spectrum estimator was developed using a multitaper technique, which yields reduced bias and variance in estimation compared to the conventional moving average method. This gives us an improved detectability measurement with finer frequency steps. The paired-view scheme in DBT reconstruction provides better image quality than the commonly used sequential method. A simple ordering like the “side-to-side” method can achieve less artifact and higher image quality in reconstructed slices. The new denoising algorithm developed was applied to the projection views acquired in DBT before reconstruction. The random noise was markedly removed while the anatomic details were maintained. With the help of this artifact-removal technique used in reconstruction and the denoising method employed on the projection views, the image quality of DBT is enhanced and lesions should be more readily detectable.

In a visual search experiment, the subject must find a target item hidden in a display of other items, and their performance is measured by their reaction time (RT). Here I look at how visual search reaction times are correlated with past reaction times. Target-absent RTs (i.e. RTs to displays that have no target) are strongly correlated with past target-absent RTs and, treated as a time series, have a 1/f power spectrum. Target-present RTs, on the other hand, are effectively uncorrelated with past RTs. A model for visual search is presented which generates search RTs with this pattern of correlations and power spectra. In the model, search is conducted by matching search items up with ¿categorizers,¿ which take a certain time to categorize each item as target or distractor; the RT is the sum of categorization times. The categorizers are drawn at random from a pool of active categorizers. After each search, some of the categorizers in the active pool are replaced with categorizers drawn from a larger population of unused categorizers. The categorizers that are not replaced are responsible for the RT correlations and the 1/f power spectrum.

The visual processing in the real world is different from the digital world. Monkey’s and a human’s visual world is richer and more colourful affording sight of flies, regardless of whether they are immobile or airborne. The study of the evolutionary process of our visual system indicates the existence of variationally spatial arrangement; from densely hexagonal in the fovea to a sparse circular structure in the peripheral retina. Normally we use a rectangular grid for the processing of images. But as per the perspective of the human eyes, the new approach is to change the grid from rectangular to hexagonal. Applying hexagonal grid in image processing is very advantageous and easy for mimicking human visual system. The main advantages for using the hexagonal structure in image processing is its resemblance to the arrangement of photoreceptors in the human eyes. The visual processing in the real world is different from the digital world. Monkey’s and a human’s visual world is richer and more colourful affording sight of flies, regardless of whether they are immobile or airborne. The study of the evolutionary process of our visual system indicates the existence of variationally spatial arrangement; from densely hexagonal in the fovea to a sparse circular structure in the peripheral retina. Normally we use a rectangular grid for the processing of images. But as per the perspective of the human eyes, the new approach is to change the grid from rectangular to hexagonal. Applying hexagonal grid in image processing is very advantageous and easy for mimicking human visual system. The main advantages for using the hexagonal structure in image processing is its resemblance to the arrangement of photoreceptors in the human eyes.

A cafeína é um potente estimulante do sistema nervoso central dos animais e vem sendo usada para melhorar o desempenho de cavalos atletas devido a sua propriedade de estímulo da atividade motora e redução da fadiga muscular. O objetivo do presente estudo foi avaliar o perfil eletroencefalográfico de equinos submetidos a cafeína comercial utilizando eletrodos de superfície e biotelemetria. Foram utilizados dois protocolos experimentais. No primeiro protocolo dois equinos (A e B) foram submetidos a cafeína comercial e no segundo protocolo dois animais controle (C e D) foram submetidos a um placebo com solução fisiológica. O EEG obtido dessas situações foi analisado no ambiente Matlab® onde se avaliou os espectros de potência. Os dados foram analisados por One-way ANOVA valores de p < 0,05 usando vários testes estatísticos. A análise do espectro resultante mostrou predominância de frequências nas faixas de 20 Hz e 35 Hz para o animal A; 15 Hz, 20 Hz e 25 Hz para o animal B, essas frequências foram verificadas nos animais antes de serem submetidos à cafeína; quando foram submetidos à cafeína foi observado um pico predominante em 10 Hz em ambos indivíduos. Para os animais controle, a frequência observada foi de 15 Hz e 25 Hz para o animal C e para o animal D as frequências foram 15 Hz, 20 Hz, 30 Hz e 35 Hz. Para ambos os animais submetidos à cafeína os resultados estatísticos comprovaram que houve diferenças entre as médias da densidade espectral de potência dos sinais adquiridos. Para os animais que foram submetidos ao placebo os testes estatísticos demonstraram que não houve diferenças das médias dos espectros constatando que a aplicação do placebo não teve efeito na atividade elétrica cerebral nos equin os estudados. Conclui-se que o EEG registrou um padrão diferenciado para os animais que foram submetidos à cafeína. / Caffeine is a powerful stimulant of the central nervous system of animals and has been used to improve performance in athletic horses due to its property of stimulating motor activity and reduced muscle fatigue. The aim of this study was to evaluate the electroencephalographic pattern of horses undergoing caffeine by the use of surface electrodes and biotelemetry system. Two experimental protocols were carried out. In the first protocol two horses (A and B) were submitted to commercial caffeine and in the second protocol two control horses (C and D) were submitted to a saline placebo. The EEG was obtained and analyzed in Matlab® by evaluating the power spectra. Data were analyzed by one-way ANOVA with p-value <0.05 using several statistical tests. The results of spectrum analysis showed predominance of frequency bands from 20 Hz to 35 Hz for animal A and 15 Hz, 20 Hz and 25 Hz for animal B; these frequencies were observed in the animals before being subjected to caffeine; when they were submitted to caffeine it was observed a predominant peak at 10 Hz in both individuals. For animals used as control the observed frequency was 15 Hz and 25 Hz for animal C; for animal D frequencies were 15 Hz, 20 Hz, 30 Hz and 35 Hz. For both animals submitted to caffeine statistical results showed that there were differences between the means of the power spectral density of signals acquired. For animals that underwent placebo saline statistical tests showed no differences of mean spectra stating that the application of placebo had no effect on brain electrical activity studied in horses. Overall results had shown that the EEG pattern was influenced by caffeine.

Philosophiae Doctor - PhD / Prevailing astronomical and astrophysical observations suggest that we live in a spatially flat cold dark matter (CDM) universe - currently going through a period of accelerated expansion possibly driven by “dark energy” in form of a cosmological constant. Within the standard cosmological paradigm, dark energy and dark matter are the dual dominant sources in the evolution of the late-time universe contributing about 70% and 25% respectively to the total energy density in the Universe, but these are only currently detected via their gravitational interaction. There could be a non-gravitational interaction within the “dark sector” without violating current observational data, thus giving rise to changes in the dark equations of state and affecting the process of galaxy formation. In this thesis, we investigate two new interesting large-scale structure formation scenarios using interacting fluids. Firstly, in departure from the standard approach in which dark matter is treated as a single independent fluid, we split the dark matter fluid into two interacting components: a strongly clustered “halo” component and a weakly clustered “free” component- accreted by the halos. By defining the fraction of the matter inside CDM “halos” to the total matter as a time evolving function of the total matter density F (ρm), we derive the governing background and perturbation equations and the energy-momentum transfer four-vectors. We then perform numerical calculations for three models for F (ρm) that are in agreement with recently published results from halo theory of N-body simulations, and compare our results to the standard ΛCDM model. Our results show that, whereas there’s a good agreement between our model and the ΛCDM model, the perturbations are much more sensitive to the interaction and can deviate strongly from the standard case for large interaction strengths. Secondly, motivated by our current poor knowledge on the underlying “dark- sector” physics and the need to understand the nature of the two most dominant components of our universe: dark energy and dark matter; we investigate a new scenario in which the two dark components interact via an energy-momentum exchange. By re-writing the evolution equations in a more suitable form, we eliminate previously reported singularities in interacting dark energy models in which dark energy is tested to be vacuum energy with w → −1. This makes it possible to numerically integrate the resulting background and perturbation equations, comparing our results to the standard model. We show that this treatment, yields a simple model that provides a good natural extension to the standard ΛCDM model. We go further to explore in detail the cosmological implications of the interaction strength and the direction of the energy-momentum transfer in vacuum interacting dark energy. This thesis provides useful insights on the possible significance of a dark sector interaction in structure formation and shows that such an interaction provides a good natural explanation for the high value of the Hubble parameters measured by BOSS and SDSS surveys. Indeed a small and positive coupling is shown to alleviate the well-known cosmological coincidence problem.

In Michelson interferometry, many algorithms are used to detect the number of active laser sources at any given time. Conventional FFT-based non-parametric methods are widely used for this purpose. However, non-parametric methods are not the only possible option to distinguish the peaks in a spectrum, as these methods are not the most suitable methods for short data records and for closely spaced wavelengths. This thesis aims to provide solutions to these problems. It puts forward the use of parametric methods such as autoregressive methods and harmonic methods, and proposes two new algorithms to detect the closely spaced peaks for different scenarios of optical signals in wavemeters. Various parametric algorithms are studied, and their performances are compared with non-parametric algorithms for different criteria, e.g. absolute levels, frequency resolution, and accuracy of peak positions. Simulations are performed on synthetic signals produced from specifications provided by our sponsor, i.e., a wavemeter manufacturing company.

In this thesis, we focus on the steady state properties of two systems which are genuinely out of equilibrium. The first project is an application of dynamic field theory to a specific non equilibrium critical phenomenon, while the second project involves both simulations and analytical calculations. The methods of field theory are used on both these projects. In the first part of this thesis, we investigate a generalization of the well-known field theory for directed percolation (DP). The DP theory is known to describe an evolving population, near extinction. We have coupled this evolving population to an environment with its own nontrivial spatio-temporal dynamics. Here, we consider the special case where the environment follows a simple relaxational (model A) dynamics. We find two marginal couplings with upper critical dimension of four, which couple the two theories in a nontrivial way. While the Wilson-Fisher fixed point remains completely unaffected, a mismatch of time scales destabilizes the usual DP fixed point. Some open questions and future work remain. In the second project, we focus on a simple particle transport model far from equilibrium, namely, the totally asymmetric simple exclusion process (TASEP). While its stationary properties are well studied, many of its dynamic features remain unexplored. Here, we focus on the power spectrum of the total particle occupancy in the system. This quantity exhibits unexpected oscillations in the low density phase. Using standard Monte Carlo simulations and analytic calculations, we probe the dependence of these oscillations on boundary effects, the system size, and the overall particle density. Our simulations are fitted to the predictions of a linearized theory for the fluctuation of the particle density. Two of the fit parameters, namely the diffusion constant and the noise strength, deviate from their naive bare values [6]. In particular, the former increases significantly with the system size. Since this behavior can only be caused by nonlinear effects, we calculate the lowest order corrections in perturbation theory. Several open questions and future work are discussed. / Ph. D.

Yes / When fixating on a stationary object, the power of the eye’s lens fluctuates. Studies have suggested that changes in these so-called microfluctuations in accommodation may be a factor in the onset and progression of short-sightedness. Like many physiological signals, the fluctuations in the power of the lens exhibit chaotic behaviour. A breakdown or reduction in chaos in physiological systems indicates stress to the system or pathology. The purpose of this study was to determine whether the chaos in fluctuations of the power of the lens changes with refractive error, i.e. how short-sighted a subject is, and/or accommodative demand, i.e. the effective distance of the object that is being viewed. Six emmetropes (EMMs, non-short-sighted), six early-onset myopes (EOMs, onset of short-sightedness before the age of 15), and six late-onset myopes (LOMs, onset of short-sightedness after the age of 15) took part in the study. Accommodative microfluctuations were measured at 22 Hz using an SRW-5000 autorefractor at accommodative demands of 1 D (dioptres), 2 D, and 3 D. Chaos theory analysis was used to determine the embedding lag, embedding dimension, limit of predictability, and Lyapunov exponent. Topological transitivity was also tested for. For comparison, the power spectrum and standard deviation were calculated for each time record. The EMMs had a statistically significant higher Lyapunov exponent than the LOMs ( 0.64±0.330.64±0.33 vs. 0.39±0.20 D/s0.39±0.20 D/s ) and a lower embedding dimension than the LOMs ( 3.28±0.463.28±0.46 vs. 3.67±0.493.67±0.49 ). There was insufficient evidence (non-significant p value) of a difference between EOMs and EMMs or EOMs and LOMs. The majority of time records were topologically transitive. There was insufficient evidence of accommodative demand having an effect. Power spectrum analysis and assessment of the standard deviation of the fluctuations failed to discern differences based on refractive error. Chaos differences in accommodation microfluctuations indicate that the control system for LOMs is under stress in comparison to EMMs. Chaos theory analysis is a more sensitive marker of changes in accommodation microfluctuations than traditional analysis methods.

The ultrasonic NDE technique to characterize the ultrasonic nonlinear response of the cyclically load 7075-T6 aluminum is described in this thesis. In order to estimate the nonlinear relation of the ultrasonic waves due to material fatigue damage or degradation, the spectral analysis techniques such as the power spectrum, bispectrum, and bicoherence spectrum are applied. The ultrasonic nonlinearity parameters by Cantrell and Jhang are introduced and presented as a function of the material fatigue growth, the number of fatigue cycles. This thesis presents the effectiveness of the bispectral analysis for evaluating the nonlinear aspects of the ultrasonic wave propagation. The results show that the nonlinearity parameters by Cantrell and Jhang are responsive to the output amplitude of the received signal and vary for the various materials, and independent of the input frequency and the ultrasonic wave propagation distance. By using the bispectral analysis tools, particularly the bicoherence spectrum, the increase of the coupling levels between the fundamental, its harmonic, and subharmonic frequency components is presented as the number of fatigue cycles is increased. This thesis suggests that the application of the bicoherence spectrum based on the nonlinear wave coupling relations be more effective for estimating the level of the material fatigue life. / Master of Science

L'objet de cette thèse est le spectre de puissance à une dimension du flux transmis dans les forêts Lyman-alpha. Les forêts Lyman-alpha sont un motif d'absorption observé dans les spectres de quasars correspondant à l'absorption de la lumière du quasar par les nuages d'hydrogène le long de la ligne de visée. C'est un outil cosmologique puissant car il sonde des échelles relativement petites de l'ordre de quelques Mpc. Il est aussi sensible à de petits effets non-linéaires tel que celui produit par des neutrinos massifs.Premièrement, nous avons développé deux méthodes indépendantes pour mesurer le spectre de puissance. La première est fondée sur une transformée de Fourier et la seconde sur une fonction de vraisemblance. Les deux méthodes sont indépendantes et ont des incertitudes systématiques différentes. La détermination du niveau de bruit dans les données spectrales a fait l'objet d'un traitement particulier, du fait de son impact significatif sur le spectre de puissance calculé. Nous avons appliqué ces méthodes à 13821 spectres de quasars provenant de la 9e publication de données de l'expérience BOSS sélectionnés à partir d'un échantillon de plus de 60000 spectres sur des critères comme le rapport signal sur bruit et la résolution spectrale. Les deux spectres de puissance mesurés sont en bon accord sur les douze domaines de décalage vers le rouge (<z>=2.2 à <z>=4.4) et sur l'ensemble des échelles (0.001 (km/s)^−1 à 0.02 (km/s)^−1). Nous avons soigneusement déterminé les incertitudes systématiques d'origine instrumentale et méthodologique de notre mesure.Ensuite, nous présentons un ensemble de simulations cosmologiques N-corps incluant de la matière noire, du gaz baryonique et des neutrinos visant à modéliser les régions de basse densité sondées par les forêts Lyman-alpha. Les simulations sont conçues pour répondre aux exigences de précision des données BOSS et eBOSS. Elles comportent 768^3 ou 192^3 particules de chaque type et explorent des volumes allant de (25 Mpc/h)^3 pour les simulations haute résolution à (100 Mpc/h)^3 pour les simulations grand volume. En utilisant une technique de raboutage, nous atteignons une précision équivalente à une simulation comportant 3072^3 particules de chaque type dans un volume de (100 Mpc/h)^3. Nous montrons que cette technique est précise à 2% sur des échelles allant de quelques Mpc jusqu'à quelques dizaines de Mpc. Nous explorons l'effet sur le spectre de puissance de 4 paramètres cosmologiques (n_s, sigma_8, Omega_m ,H_0), 2 paramètres astrophysiques (T_0, gamma) décrivant la relation température/densité du milieu intergalactique et de la somme des masses des neutrinos. En faisant varier ces paramètres autour d'un modèle central choisi en accord avec les résultats de Planck, nous avons construit une grille de simulations, permettant non seulement l'étude de l'effet de chaque paramètre individuellement mais aussi l'effet de chaque paire de paramètres. Nous obtenons ainsi un développement au deuxième ordre complet, incluant les termes croisés, autour de notre modèle central. Nous avons vérifié la validité de ce développement avec des simulations indépendantes obtenues soit avec des paramètres différents soit une graine différente pour la génération des conditions initiales. Une comparaison entre le spectre de puissance mesuré à partir des données dans la première partie et celui obtenu à partir de nos simulations montre un excellent accord.Enfin, même s'il reste des biais potentiels et des erreurs systématiques à étudier dans nos simulations, nous avons réalisé des ajustements en combinant notre mesure du spectre de puissance à d'autres sondes cosmologiques comme les mesures du fond diffus cosmologique par le satellite Planck. Ces résultats préliminaires sont très encourageants car ils mènent à des contraintes sur les paramètres cosmologiques parmi les plus précises à ce jour, en particulier sur la sommes des masses des neutrinos avec une limite supérieure à 0.1 ev. / The framework of the studies presented in this thesis is the one-dimensional power spectrum of the transmitted flux in the Lyman-alpha forests. The Lyman-alpha forest is an an absorption pattern seen in the spectra of high redshift quasars corresponding to the absorption of the quasar light by the hydrogen clouds along the line of sight. It is a powerful cosmological tool as it probes relatively small scales, of the order of a few Mpc. It is also sensible to small non-linear effects such as the one induced by massive neutrinos.First, we have developed two independent methods to measure the one-dimensional power spectrum of the transmitted flux in the Lyman-alpha forest. The first method is based on a Fourier transform, and the second on a maximum likelihood estimator. The two methods are independent and have different systematic uncertainties. The determination of the noise level in the data spectra was subject to a novel treatment, because of its significant impact on the derived power spectrum. We applied the two methods to 13,821 quasar spectra from SDSS-III/BOSS DR9 selected from a larger sample of over 60,000 spectra on the basis of their high quality, large signal-to-noise ratio, and good spectral resolution. The power spectra measured using either approach are in good agreement over all twelve redshift bins from <z>=2.2 to <z>=4.4, and scales from 0.001 (km/s)^−1 to 0.02 (km/s)^−1. We carefully determined the methodological and instrumental systematic uncertainties of our measurements.Then, we present a suite of cosmological N-body simulations with cold dark matter, baryons and neutrinos aiming at modeling the low-density regions of the IGM as probed by the Lyman-alpha forests at high redshift. The simulations are designed to match the requirements imposed by the quality of BOSS and eBOSS data. They are made using either 768^3 or 192^3 particles of each type, spanning volumes ranging from (25 Mpc/h)^3 for high-resolution simulations to (100 Mpc/h)^3 for large-volume ones. Using a splicing technique, the resolution is further enhanced to reach the equivalent of simulations with 3072^3 = 29 billion particles of each type in a (100 Mpc/h)^3 box size, i.e. a mean mass per gas particle of 1.2x10^5 solar masses. We show that the resulting power spectrum is accurate at the 2% level over the full range from a few Mpc to several tens of Mpc. We explore the effect on the one-dimensional transmitted-flux power spectrum of 4 cosmological parameters (n_s, sigma_8, Omega_m ,H_0), 2 astrophysical parameters (T_0, gamma) related to the heating rate of the IGM and the sum of the neutrino masses. By varying the input parameters around a central model chosen to be in agreement with the latest Planck results, we built a grid of simulations that allows the study of the impact on the flux power spectrum of these seven relevant parameters. We improve upon previous studies by not only measuring the effect of each parameter individually, but also probing the impact of the simultaneous variation of each pair of parameters. We thus provide a full second-order expansion, including cross-terms, around our central model. We check the validity of the second-order expansion with independent simulations obtained either with different cosmological parameters or different seeds for the initial condition generation. Finally, a comparison to the one-dimensional Lyman-alpha forest power spectrum obtained in the first part with BOSS data shows an excellent agreement.Eventually, even if there are still some potential biases and systematic errors that need to be studied in our simulation, we performed cosmological fits combining our measurement of the one-dimensional power spectrum and other cosmological probes such as the CMB results provided by Planck. These preliminary results are very encouraging as they lead to some of the tighest cosmological constraints as of today, especially on the sum of the neutrino masses with an upper limit of 0.1 eV.