The simultaneous acquisitions of electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) have been applied to improve the surgery planning of patients with drug resistant epilepsy. However, the classical approach of analyzing the EEG-fMRI data is inefficient in patients whom only few or non interictal epileptiforms discharges (IEDs) are detected during the simultaneous acquisition. Another issue of EEG-fMRI acquisition is related to its high sensitivity to motion, which decreases the quality of both data, even worse in non-cooperative patients. In this work we propose and discuss the application of two methods of analyzing fMRI data of patients with focal epilepsy: Independent component analysis (ICA) and two-dimensional temporal clustering (2dTCA). Each method was applied in a distinct group of patients and the results were compared to those obtained by the classic EEG-fMRI analysis. We have also proposed a method to improve the quality of EEG data using the head position measurements obtained, by a prospective motion correction (PMC) system, during the EEG-fMRI acquisitions. In the ICA study, we have used the electrical source images for selecting independent components (ICs) in EEG data of 13 patients with different spiking frequency. The method detected epilepsy-related BOLD activity in all the patients. Comparatively, the classic EEG-fMRI could be applied in 11 patients and epilepsy-related BOLD activities were found in seven of them. In the 2dTCA study, we have evaluated 20 patients and found epilepsy-related maps in 14 of them. Thirteen of the twenty patients have IEDs detected during the simultaneous acquisition; the classic EEG-fMRI provided maps related to the epileptogenic region in six of them. Finally we have verified in three health subjects that the proposed method for correcting motion-induced artefacts in the EEG data is effective for high amplitude and velocities (~1cm and 55mm/s). We concluded that the ICA and 2dTCA methods increase the sensitivity of using fMRI for mapping the epileptogenic region, mainly in patients presenting few or no IEDs in the EEG data simultaneously acquired to the fMRI. The PMC use during the fMRI acquisition does not degrade the quality of the EEG data acquired simultaneously. In fact, the motion information can be used for improving its quality by correcting motion-induced artefacts. / As aquisições simultâneas de dados de eletroencefalografia (EEG) e imagens funcionais por ressonância magnética (fMRI) vêm sendo utilizadas com intuito de melhorar o planejamento cirúrgico de pacientes com epilepsia refratária. Entretanto, o processamento classicamente usado nestes dados combinados não é possível em pacientes sem descargas epileptiformes interictais (IEDs) e possui baixa sensibilidade para aqueles em que poucas IEDs são detectadas durante a aquisição simultânea. Além disto, a técnica é sensível ao movimento dos pacientes durante as aquisições, o que reduz a qualidade dos dados, principalmente em pacientes não cooperantes. Neste trabalho é proposto e discutido o uso de dois métodos de processamento, baseados nas técnicas de análise de componentes independentes (ICA) e análise temporal de clusters em duas dimensões (2dtca), para se mapear regiões epileptogênicas. Cada método foi analisado em um conjunto diferente de pacientes e os resultados foram comparados com os obtidos pelo EEG-fMRI clássico. Finalmente, propomos um método que utiliza às medidas de posicionamento da cabeça, obtidas durante a aquisição das fMRI, para aumentar a qualidade dos dados de EEG adquiridos simultaneamente. No estudo usando ICA combinado com imagens de fontes elétricas analisamos os dados de 13 pacientes com diferentes frequências de descargas e observamos que este método encontrou ao menos uma componente independente relacionada à epilepsia em cada paciente. Comparativamente usando o processamento convencional foi possível avaliar 11 dos 13 pacientes, e em apenas sete deles os mapas resultantes foram considerados concordantes com a região epileptogênica (RE). No estudo utilizando 2dTCA avaliamos 20 pacientes e encontramos mapas relacionados com a RE em 14 deles. Neste conjunto de pacientes, 13 apresentaram IEDs durante as aquisições; neles o método clássico de processamento teve resultados concordantes com a RE em seis deles. Finalmente verificamos em três sujeitos saudáveis que o método aqui proposto para corrigir os artefatos induzidos no EEG devido ao movimento é efetivo para altas amplitudes e velocidades (~1cm e 55mm/s). Concluímos que os métodos ICA e 2dTCA aumentam a sensibilidade do uso de fMRI para mapear RE, principalmente em pacientes com baixa ou nenhuma detecção de IEDs durante às aquisições. Também concluímos que o uso da correção prospectiva de movimento em aquisições de fMRI não reduz a qualidade do dado de EEG adquirido simultaneamente e que às informações de movimento mensuradas podem melhorar a qualidade deste dado em situações de repouso e movimento do sujeito durante o experimento.

O mapeamento das respostas BOLD (Blood Oxygenation Level Dependent) constitui etapa importante nos experimentos de imagem funcional por ressonância magnética (Functional Magnetic Resonance Imaging fMRI) e de EEG-fMRI simultâneo. Em sua grande maioria, a análise de dados de fMRI e de EEG-fMRI está baseada no modelo linear geral (General Linear Model GLM), que procura localizar as respostas BOLD por meio de modelos definidos a priori. Porém, em muitos casos, como em pacientes, variações na forma e/ou atraso podem reduzir a confiabilidade dos resultados. Desse modo, o primeiro objetivo deste trabalho foi explorar métodos clássicos e propor novos métodos para análise de dados de fMRI e de EEG-fMRI simultâneo. Neste trabalho, um método modificado baseado na distância de Kullback-Leibler generalizada (dKLg) foi desenvolvido. Diferentemente do GLM, essa abordagem não requer um modelo para a resposta. Dados simulados foram utilizados para otimizá-lo e compará-lo ao GLM sob diferentes condições de resposta como a relação sinal ruído e a latência. Em seguida, o dKLg foi testado em dados reais, adquiridos em 14 voluntários assintomáticos, submetidos a tarefas motoras e auditivas padrões. Os resultados mostram a equivalência entre o dKLg e o GLM. Em seguida, essa estratégia foi testada em 02 pacientes com com estenose carotídea unilateral. Neste caso, o dKLg foi capaz de detectar regiões significativas ipsilaterais à estenose, não detectadas pelo GLM, em virtude do atraso do sinal BOLD. Em seguida, esses métodos foram aplicados sobre exames de EEG-fMRI realizados em 45 pacientes com epilepsia. Para esse conjunto de dados, mais uma abordagem foi elaborada, que utiliza a Análise de Componentes Independentes (Independent Component Analysis ICA). Denominado ICA-GLM, ele permite extrair de modo semi-automático a amplitude, duração e topografia das descargas epileptiformes interictais (Interictal Epileptiform Discharges IED), favorecendo a inclusão de sinais do EEG de menor destaque. Além dessa vantagem, ele ainda permite incluir modelos do sinal BOLD com diferentes latências, aumentando a abrangência da variabilidade das respostas encontradas em pacientes com epilepsia. A eficiência do ICA-GLM também foi comparado à do GLM e dKLg nos exames de EEG-fMRI. Embora os resultados tenham demonstrado a robustez do GLM, em alguns pacientes o dKLg foi mais eficiente para localizar regiões concordantes que não foram detectadas pelo GLM. Ainda, em boa parte dos casos o ICA-GLM detectou regiões mais extensas e com maior valor estatístico, quando comparado ao GLM. De forma geral, nota-se que o dKLg e ICA-GLM podem ser ferramentas complementares importantes ao GLM, aumentando a sensibilidade dos exames de EEG-fMRI como um todo. Outra etapa importante nas avaliações de EEG-fMRI em pacientes com epilepsia tem sido a utilização de imagens de fontes elétricas (Electrical Source Imaging ESI). Neste trabalho, os mapas de ESI foram obtidos por dois métodos de solução inversa distribuída nunca usados no cenários da EEG-fMRI: Bayesian Model Averaging (BMA) e constrained Low Resolution Electromagnetic Tomography (cLORETA). Além da construção dos mapas de ESI, avaliamos a utilidade de combinar as técnicas de ESI e de EEG-fMRI para promover a diferenciação entre fontes primárias e de propagação temporal. Essa análise permitiu avaliar a concordância entre as regiões detectadas pelo ESI e EEG-fMRI e diferenciar as respostas BOLD relacionadas aos componentes iniciais e posteriores da IED. Embora os resultados ainda sejam preliminares para eleger qual método seria mais eficiente (cLORETA ou BMA), a distância encontrada entre o máximo ESI e o cluster de EEG-fMRI mais próximo foi consistentemente similar, em ambos, com os dados recentes da literatura. / Functional magnetic resonance imaging (fMRI) and combined EEG-fMRI usually rely on the successful detection of Blood Oxygenation Level Dependent (BOLD) signal. Typically, the analysis of both fMRI and EEG-fMRI are based on the General Linear Model (GLM) that aims at localizing the BOLD responses associated to an a priori model. However, the responses are not always canonical, as is the case of those from patients, which may reduce the reliability of the results. Therefore, the first objective of the present study was to explore the usage of classical methods, such as the GLM, and to propose alternative strategies to the analysis of fMRI and combined EEG-fMRI. A first method developed was based on the computation of the generalized Kullback-Leibler distance (gKLd), which does not require the use of an a priori model. Simulated data was used to allow quantitative comparison between the gKLd and GLM under different response conditions such as the signal to noise ratio and delay. The gKLd was then tested on real data, first from 14 asymptomatic subjects, submitted to classical motor and auditory fMRI protocols. The results demonstrate that under these conditions the GLM and gKLd are equivalent. The same strategy was applied to 02 patients with unilateral carotid stenosis. Now the dKLg was capable of detecting the expected bilateral BOLD responses that were not detected by the GLM, as a consequence of the response delay imposed by the stenosis. Those comparisons were now extended to the evaluation of EEG-fMRI exams from 45 patients with epilepsy. For this data set, an additional method was used, based on the use of Independent Component Analysis (ICA), which was called ICA-GLM. It allows extracting semi-automatically the amplitude, duration and topography of EEG interictal Epileptiform Discharges (IED), favoring the use of less prominent signals. Moreover, it also allows the use of BOLD response models with different delays, expanding the variability of the responses to be detected in patients with epilepsy. ICA-GLM was also compared to GLM and dKLg in these EEG-fMRI evaluations. Although in general the results have demonstrated the robustness of the GLM, dKLg was more efficient in detecting the responses from some pacients, while the ICA-GLM mostly detected broader regions with more significant results when compared to GLM. In general, dKLg and ICA-GLM seem to offer an important complementary aspect to the GLM, increasing its sensibility in EEG-fMRI as a whole. Another important aspect of EEG-fMRI applied to patients with epilepsy has been the inspection of Electrical Source Imaging (ESI) to evaluate some dynamical aspects of the IED. Herein, ESI maps were obtained from two inverse distributed solutions that were not applied so far to EEG-fMRI: Bayesian Model Averaging (BMA) and constrained Low Resolution Electromagnetic Tomography (cLORETA). Besides, we also evaluated the combined information from ESI and EEG-fMRI in order to differentiate from primary sources to temporal propagation of the signal. Such analysis allowed us to inspect for the correspondence between regions detected by ESI e EEG-fMRI and to separate BOLD signals whose sources are related to the initial and later components of the IED. Although the results are preliminary to determine which ESI method (cLORETA or BMA) is more efficient, the distance between the maximum ESI and the closest EEG-fMRI cluster was consistently similar with those reported in the literature.

" Electroencephalogram (EEG), Electrocardiogram (ECG), and Electromyogram (EMG) bio-potential signals are commonly recorded in clinical practice. Typically, patients are connected to a bulky and mains-powered instrument, which reduces their mobility and creates discomfort. This limits the acquisition time, prevents the continuous monitoring of patients, and can affect the diagnosis of illness. Therefore, there is a great demand for low-power, small-size, and ambulatory bio-potential signal acquisition systems. Recent work on instrumentation amplifier design for bio-potential signals can be broadly classified as using one or both of two popular techniques: In the first, an AC-coupled signal path with a MOS-Bipolar pseudo resistor is used to obtain a low-frequency cutoff that passes the signal of interest while rejecting large dc offsets. In the second, a chopper stabilization technique is designed to reduce 1/f noise at low frequencies. However, both of these existing techniques lack control of low-frequency cutoff. This thesis presents the design of a mixed- signal integrated circuit (IC) prototype to provide complete, programmable analog signal conditioning and analog-to-digital conversion of an electrophysiologic signal. A front-end amplifier is designed with low input referred noise of 1 uVrms, and common mode rejection ratio 102 dB. A novel second order sigma-delta analog- to-digital converter (ADC) with a feedback integrator from the sigma-delta output is presented to program the low-frequency cutoff, and to enable wide input common mode range of ¡Ãƒâ€œ0.3 V. The overall system is implemented in Jazz Semiconductor 0.18 um CMOS technology with power consumption 5.8 mW from ¡Ãƒâ€œ0.9V power supplies. "

"A combined hardware and software platform for ambulatory seizure onset detection is presented. The hardware is developed around commercial off-the-shelf components, featuring ADS1299 analog front ends for electroencephalography from Texas Instruments and a Broadcom ARM11 microcontroller for algorithm execution. The onset detection algorithm is a patient-specific support vector machine algorithm. It outperforms a state-of-the-art detector on a reference data set, with 100% sensitivity, 3.4 second average onset detection latency, and on average 1 false positive per 24 hours. The more comprehensive European Epilepsy Database is then evaluated, which highlights several real-world challenges for seizure onset detection, resulting in reduced average sensitivity of 93.5%, 5 second average onset detection latency, and 85.5% specificity. Algorithm enhancements to improve this reduced performance are proposed."

EEG has been used to explore the electrical activity of the brain for manydecades. During that time, different components of the EEG signal have been iso-lated, characterized, and associated with a variety of brain activities. However, nowidely accepted model characterizing the spatio-temporal structure of the full-brainEEG signal exists to date.Modeling the spatio-temporal nature of the EEG signal is a daunting task. Thespatial component of EEG is defined by the locations of recording electrodes (rang-ing between 2 to 256 in number) placed on the scalp, while its temporal componentis defined by the electrical potentials the electrodes detect. The EEG signal is gen-erated by the composite electrical activity of large neuron assemblies in the brain.These neuronal units often perform independent tasks, giving the EEG signal ahighly dynamic and non-linear character. These characteristics make the raw EEGsignal challenging to work with. Thus, most research focuses on extracting andisolating targeted spatial and temporal components of interest. While componentisolation strategies like independent component analysis are useful, their effective-ness is limited by noise contamination and poor reproducibility. These drawbacks tofeature extraction could be improved significantly if they were informed by a globalspatio-temporal model of EEG data.The aim of this thesis is to introduce a novel data-surface reconstruction (DSR)technique for EEG which can model the integrated spatio-temporal structure of EEGdata. To produce physically intuitive results, we utilize a hyper-coordinate transfor-mation which integrates both spatial and temporal information of the EEG signalinto a unified coordinate system. We then apply a non-uniform rational B spline(NURBS) fitting technique which minimizes the point distance from the computedsurface to each element of the transformed data. To validate the effectiveness of thisproposed method, we conduct an evaluation using a 5-state classification problem;with 1 baseline and 4 meditation states comparing the classification accuracy usingthe raw EEG data versus the surface reconstructed data in the broadband rangeand the alpha, beta, delta, gamma and higher gamma frequencies. Results demon-strate that the fitted data consistently outperforms the raw data in the broadbandspectrum and all frequency spectrums.

This thesis set out to investigate the electrophysiological correlates of maths anxiety (MA). Research has shown that those with high MA (HMA) tend to have poorer accuracy and increased reaction time on maths based tasks and that high maths anxious individuals avoid situations where they might have to use maths. This can impact on their future by restricting their degree or job prospects. Previous research has identified the behavioural cognitive and psychological effects of MA and recently studies have begun to examine the associated underlying mechanisms in the brain. Chapter one outlines the background MA behavioural and measurement research before evaluating the neurophysiological methods used in cognitive neuroscience and the use of electroencephalography (EEG) in chapter two. Chapter three continues by outlining previous research concerning the neurophysiological processing of maths and number before evaluating relevant neurophysiological research concerning MA. Four experimental studies are conducted, exploring the neurophysiological underpinnings of MA research using EEG. Each of these recruits 30 participants and measures of electro-cortical (Event Related Potentials (ERPs), Global Field Power, Frequency etc.) and questionnaire measures are implemented. The first study aimed to identify whether the behavioural effects of MA (poorer accuracy and increased reaction time) are consistent with ERP differences (component amplitude and latency differences) in the brain and to understand why these effects are experienced. This revealed no significant comparisons between ERP components and behavioural responses involving low and high maths anxious individuals, but this may have been due to the lack of an anxious response by using a verification task, rather than requiring calculation. Study two introduces the measurement of gamma activity as a neurophysiological measure of anxiety and threat processing and brings three core areas of anxiety research together: Previous studies outline high anxiety in connection with gamma modulation, also showing gamma band activity is associated with the amygdala and finally, that the amygdala is responsible for the processing of threat perception and anxiety. This research has not been brought together when studying MA. Results produced similar ERP findings to the previous study but the introduction of gamma activity into the research provided the first differences between high and low MA (LMA) groups, showing significantly greater gamma activity levels in HMA individuals. However, this study only used numerically-based tasks, thus the third study implemented a non-numerical condition to act as a control. Study three replicates the findings showing a reduced level of gamma activity in high MA individuals for the non-numerical based task, however, this was also reduced for the simple maths task. It was theorised that it is more likely to be the initial threat perception that represents the anxious response and gamma activity increases. To test this and remove any working memory demands, the fourth study implements the presentation of single digit observation (using single digit numbers and letters). Even though there was no demand on working memory, high maths anxious participants displayed similar levels of gamma activity as low maths anxious individuals during letter observation. However, they had significantly greater levels during the observation of number. Findings suggest that HMA individuals may not only struggle with the processing of maths stimuli, but may have a threat-related response to the simple observation of numerical stimuli. This implies that HMA individuals consistently apply an avoidance technique due to a threat response associated with increased levels of gamma activity. The findings of the each study are finally discussed in terms of their contribution to the neurophysiological underpinnings of MA, the first exploration of this using gamma activity, future research and the extent that number anxiety may act as a precursor or sine qua non to MA.

Multiple elements of the retail environment can have an impact on a consumer's behaviour and purchase decisions. Much of the influence that the environment has on behaviour often goes unnoticed, as it affects internal processes that happen below the level of conscious awareness. This research aims to explore and quantify the effect a retail environment has on consumers' affective (emotional) and cognitive responses towards products. Priming is the influence of external stimuli on one's behaviour or response towards target stimuli. This research designed an experiment to prime participants with a particular coloured stimulus (pink, blue, or red) in order to measure the influence of this prime on the participants' purchase decisions. The participants entered a real-world simulated retail shop, and within a guided format they shopped through the available dresses, eventually picking out their three ranked favourites. The participants' physiological responses were measured using an eye-tracker and a portable Electroencephalogram (EEG) recording unit. The eye-tracking data were analysed using the Gaze Cascade Theory, testing for an increase in gaze bias towards preferred and primed products. The EEG data provided information about the participants' brain activity, and were analysed in accordance with Davidson's model of emotion, indicating an approach or withdrawal tendency towards different products. The results showed that with both eye-tracking and EEG it is possible to measure a difference between the participants' cognitive and affective responses towards the products that they preferred and chose as their favourites, compared with the products they did not choose. The EEG data provided evidence of a difference in neural responses between the prime matching coloured products and the non-prime matching products. However, the eye-tracking responses did not demonstrate a significant difference in eye-movements between the primed and not primed products. Technical innovation was required to allow the recording of EEG data in the semi-controlled shop environment, to allow data free of motion artefacts to be analysed. These results demonstrate the ability to measure consumers' physiological, neural, and subconscious responses in a real-world retail environment, whilst allowing the participants to move freely and unhindered. A novel methodology for analysing motion artefact free EEG data is presented. The results demonstrate a significant difference in emotional responses, as detected by EEG, in preference towards the prime coloured products, suggesting that priming has an influence in decision making in fashion retail environments.

Navigation through the environment requires the ability to select relevant information from a multitude of irrelevant stimuli. Under conditions of processing conflict, attention and cognitive control processes bias sensory input based on internal goals. These processes are supported by the interplay of a fronto-parietal attention network that exerts a top-down influence on information processing and a superior temporal network that operates in parallel and that responds in a stimulus-driven manner to behaviorally salient stimuli. It is often reported that nicotine can enhance top-down attention control and reduce distraction. In experiments 1 and 2, the effects of increasing control demands on behavior were assessed using electrophysiological (EEG) and behavioral measures in an auditory number parity decision task with different levels of distraction. Participants made forced choice ‘odd’ or ‘even’ number decisions, while ignoring preceding or simultaneous novel distractors. A group of non-smokers was compared to overnight abstinent smokers (9 hours) and after nicotine intake via 2 mg nicotine tablet or via smoke-inhaled nicotine. The results revealed that preceding distractors impaired task performance due to orienting to and reorienting from the distractor. Simultaneous distractors did not cause orientation of attention (indicated by absence of a P3a Event-Related Potential) and produced smaller increments in response latencies. However, this type of complex novel stimulus initiated processes of memory updating that significantly impaired response sensitivity and accuracy. Nicotine withdrawal enhanced these distraction effects, whereas nicotine intake, particularly via smoking, normalized performance. In experiment 3, dichotic listening performance in a group of non-smokers was compared to abstinent smokers (12 hours) using behavioral, EEG and functional Magnetic Resonance Imaging (fMRI) measures. The perceptual salience of the stimuli was manipulated by systematically varying the Inter-aural Intensity Difference (IID) between them. The analysis pointed to distinct brain networks that differentially activate depending on the level of competition between sensory inputs and these effects were additionally modulated by nicotine withdrawal. Nicotine withdrawal impaired behavioral performance supported by evidence of enhanced use of memory and attention resources, and some evidence of task-independent default mode network activation. Overall, the findings suggest that withdrawal from nicotine, particularly in heavy smokers, is associated with impairments in cognitive control and that subsequent intake of nicotine serves mainly to normalize performance.

Brain Computer Interface (BCI) technology can be important for those unable to communicate due loss of muscle control. The P300 Speller allows communication at a rate up to eight selections per minute. Given this relatively slow rate of communication highly accurate classification is of great importance. Previous studies have shown that alternative stimuli (e.g., faces) can improve BCI speed and accuracy. The present study uses two new alternative stimuli, locations and graspable tools in a two-stimulus paradigm. Functional MRI studies have shown that images of familiar locations produce brain responses in the parahippocampal place area and graspable tools produce brain responses in premotor cortex.The current study shows that location and tool stimuli produce unique brain responses that can be used for classification in the two-stimulus paradigm. This study shows proof of concept for using two unique stimuli to improve speed and accuracy of the P300 Speller.

Medical choices can be life or death, and thus improving the accuracy of diagnostic decisions within a time constrained environment has a large potential for positive change. To that end, an adaptation of Dual Process Theory was developed to create a theoretical framework for medical decision making. In order to effectively measure this framework, a possible electroencephalographical link was investigated. During a complex medical diagnostic task, 52 participants were asked to diagnose what liver condition simulated patients had based on procedurally generated biometric data Feedback was provided during a learning phase until the pattern was learned. During the experimental phase, possible ranges for the biometric data were extended, allowing for increased diagnostic difficulty in some trials, thereby producing conflict for the participants. This difference between the control (Type 1) trials and the high conflict (Type 2) trials was measured using electroencephalography. It was predicted that an elevation in midfrontal theta power would be observed in high-conflict trials, which would provide a neurological correlate for Type 2 processing. This hypothesis was not verified, although several modifications to the experimental design were provided to inform future investigations. It is likely that an improved paradigm would be able to distinguish between the two processes, providing vital neurofeedback that could inform future medical students and emphasize effective learning to improve diagnostic outcomes. / Graduate