O dito popular afirma que quando uma pessoa perde um dos sentidos há uma compensação por parte dos outros sentidos para suprir a perda. Através de três experimentos psicofísicos baseados no modelo de Posner (Inibição de Retorno) e técnicas de Ressonância Magnética Funcional, surdos congênitos foram comparados a pessoas ouvintes com o objetivo de verificar se os surdos possuem processos atencionais diferentes dos ouvintes, e se as mesmas áreas corticais como a área de Wernicke, Broca e Córtex auditivo - eram ativadas em ambos os grupos. A tarefa consistia em pressionar um botão todas as vezes que os sujeitos detectassem a presença de um quadrado maior (alvo) apresentado em uma tela, enquanto também eram apresentados quadrados menores (pista) ora do mesmo lado, ora do lado oposto ao alvo. Através do Experimento I se pôde verificar que ambos os grupos apresentaram os fenômenos clássicos do Paradigma de Posner: Facilitação ou Inibição de Retorno, o que denotou a possibilidade de mecanismos atencionais semelhantes para ambos os grupos. Foi observado, porém, que os ouvintes eram mais rápidos que os surdos para responder à tarefa quando o intervalo temporal entre pista e alvo era longo (800ms), comparado ao tempo que levavam para responder quando o intervalo entre pista e alvo era curto (100 ms). O Experimento I suscitou a hipótese de que os surdos possivelmente apresentassem uma diferença de processamento temporal. No Experimento I todas as condições eram apresentadas de forma randômica. O Experimento II foi elaborado com o objetivo de por em evidência a 22 diferença dos TRM para intervalos curtos e longos, portanto os intervalos entre pista e alvo passaram a ser apresentados de forma fixa. Ao comparar os resultados do Experimento I com os do Experimento II (Intervalos Temporais Fixos), se pode verificar que os ouvintes apresentaram Tempos de Reação Manual mais lentos, enquanto os surdos apresentaram as mesmas médias a despeito da vantagem temporal, o que levou a sugerir a hipótese de que os surdos apresentem um déficit no processamento temporal. O experimento III consistiu na utilização do Paradigma de Posner enquanto os sujeitos eram submetidos ao exame de Ressonância Magnética Funcional com o objetivo de investigar se as regiões corticais ativadas poderiam ser semelhantes nos dois grupos. As imagens por Ressonância Magnética Funcional (RMF) demonstraram ativações nas áreas de Wernickie, Broca, e córtex auditivo em ambos os grupos enquanto executavam a tarefa, que embora não possuísse nenhum contexto semântico explícito, possuía o tempo como o principal parâmetro físico no qual os sujeitos pudessem se basear para melhorar o desempenho na tarefa. O tempo é um dos parâmetros físicos primários da língua oral, diferente da língua de sinais que possui o parâmetro visual e espacial como primário. Os resultados sugerem que as ativações corticais nos centros de audição e fala podem indicar uma plasticidade cruzada no grupo de surdos. Ainda, a participação do córtex auditivo no processamento da elaboração de estratégias para responder a uma tarefa que não contenha um contexto semântico explicito possivelmente indica sua participação no processamento de linguagem. / It is popularly said that when a person loses one sense, there is a compensation by the other remaining senses to suppress the loss. Throughout three Phsycophysic Experiments based on Inhibition of Return Posners Paradigm and Functional Magnetic Resonance (fMRI) Techniques, congenital deaf people were compared to normal hearing people in order to check if deaf people possess different attentional pattern compared to normal hearing people, and if the same cortical areas Wernicke and Brocas area and Hearing Cortex were activated in both groups. Experiment I consisted on pressing a button every time the presence of a big square (target) was detected by subjects while non-predictive small squares (cue) were also presented at the same or opposite side of the target. At Experiment I it was observed that both groups presented Posners Paradigm classical phenomena: Facilitation or Inhibition of Return, what suggested the possibility that attentional pattern may be similar to both groups. Therefore, it was observed that normal hearing people were faster than deaf people to respond to the task when time interval between cue and target was long (800 ms) when compared to the time they spent to respond when time interval between cue and target as short (100 ms). 24 Experiment I raised the hypotheses that possibly deaf people may present a temporal processing difference. At Experiment I every condition was randomly presented. Experiment II was elaborated to highlight MRT differences between short and long time intervals, so every time interval was presented on a fixed order. Comparison of Experiment I and II (Fixed Time Intervals) showed that normal hearing people presented shorter Manual Reaction Times (MRT), while deaf people kept the same averages despite the temporal advantage, what suggested that deaf people may present a deficit on temporal processing. Experiment III used Posners Paradigm while subjects were submitted to fMRI scanning in order to check if activated cortical regions could be similar in both groups. fMRI images demonstrate Wernicke and Brocas area and hearing cortex activations in both groups while executing the task, which, although did not have any explicit semantic content, had time as the main physical parameter on which subjects could be based to increase performance to respond to the task. Time is one of the oral language primary physical parameter, different of signed language which has visual and spatial parameters as primaries. Results suggest that cortical audition center activations may indicate a cross-modal plasticity at the deaf group. Yet, participation of hearing cortex on strategy elaboration to respond to a task which does not have any explicit semantic content possibly indicates the participation of hearing cortex on language processing.

Esta dissertação desenvolve e aplica métodos para caracterizar regiões cerebrais durante o estado de repouso. Utilizam-se grafos para representar a inter-dependência temporal de sinais de ressonância magnética funcional provenientes de regiões cerebrais distintas. Vértices representam regiões cerebrais e arestas representam a conectividade funcional. Buscando superar os problemas de visualização e interpretação desta forma de representação, elaboram-se métodos quantitativos para caracterizar padrões de conectividade entre regiões cerebrais. Para cada sujeito analisado: 1) Faz-se a redução da dimensionalidade espacial das imagens de ressonância magnética funcional respeitando os limites anatômicos das regiões cerebrais. 2) Estima-se a rede de conectividade funcional pela coerência direcionada entre pares de regiões distintas. 3) Constrói-se um grafo direcionado e pesado pela medida de conectividade. 4) Quantificam-se os vértices por índices e faz-se o registro destes valores no espaço comum MNI. 5) Avalia-se a consistência de cada índice pelo teste não paramétrico de Friedman seguido de análises de múltiplas comparações. A análise de 198 imagens de sujeitos sadios produziu resultados consistentes e biologicamente plausíveis. Em sua maioria, revelou regiões associadas a conceitos anatômicos de conectividade e integração cerebral. Embora de implementação simples, o método proporciona informações de natureza dinâmica sobre as relações entre diferentes regiões cerebrais e pode ser utilizado futuramente para estudar e entender desordens psiquiátricas/neurológicas. / This dissertation develops and applies methods to characterize brain regions during resting state. Graphs are used to represent functional MRI connectivity from different brain regions. Vertices represent brain regions and edges represent connectivity. To overcome the visualization and interpretation problems of this form of representation, we developed quantitative methods to characterize its patterns. Methods: For each subject: 1) The reduction of spatial dimensionality of functional magnetic resonance imaging is carried out taking into account the anatomic limits of the brain regions. 2) The network is estimated by directed coherence between pairs of separate regions. 3) A directed graph with weights on its edges is constructed using the later connectivity measure. 4) The vertices are quantified by indexes that are registered in the MNI common space. 5) The consistency of each index is evaluated by the nonparametric Friedman followed by Post-Hoc analysis. Results: The analysis of 198 images of healthy subjects produced consistent and biologically plausible results. They revealed anatomical regions involved in brain integration. Conclusion: The method provides information about the dynamic nature of the relationships between different brain regions and can be used in future clinical studies to understand psychiatric and neurological disorders.

BACKGROUND: Social cognitive ability is a significant determinant of functional outcome and deficits in social cognition are a disabling symptom of psychotic disorders. The neurobiological underpinnings of social cognition are not well understood, hampering our ability to ameliorate these deficits. Using ‘resting-state’ fMRI (functional magnetic resonance imaging) and a trans-diagnostic, data-driven analytic strategy, we sought to identify the brain network basis of emotional intelligence, a key domain of social cognition. METHODS: Subjects included 60 participants with a diagnosis of schizophrenia or schizoaffective disorder and 46 healthy comparison participants from three different sites: Beth Israel Deaconess Medical Center, Boston, MA, McLean Hospital, Belmont, MA, and University of Pittsburgh, Pittsburgh, PA. All participants underwent a structural T1/MPRAGE and resting-state fMRI scan. Emotional Intelligence was measured using the Mayer-Salovey-Caruso Emotional Intelligence Test (MSCEIT). A connectome-wide analysis of brain connectivity examined how each individual brain voxel’s connectivity correlated with emotional intelligence using multivariate distance matrix regression (MDMR). RESULTS: We identified a region in the left superior parietal lobule (SPL) where individual network topology predicted emotional intelligence. Specifically, the association of this region with the Default Mode Network (DMN) predicted higher emotional intelligence (r = 0.424, p < 0.001) and association with the Dorsal Attention Network (DAN) predicted lower emotional intelligence (r = -0.504, p < 0.001). This correlation was observed in both schizophrenia and healthy comparison participants. These results held true despite corrections for sex, age, race, medication dosage (chlorpromazine equivalents), and full scale IQ (FSIQ), and was replicable per site. Post-hoc analyses showed that membership of the left SPL was entirely within the DMN in high scorers and within the DAN in low scorers. This relationship was also shown to be specific to the identified left SPL region when compared to adjacent regions. Sulcal depth analysis of the left SPL revealed a correlation to emotional intelligence (r = 0.269, p = 0.0075). CONCLUSIONS: Previous studies have demonstrated individual variance in brain network topology but the cognitive or behavioral relevance of these differences was undetermined. We observe that the left SPL, a region of high individual variance at the cytoarchitectonic level, also demonstrates individual variance in its association with large scale brain networks and that network topology predicts emotional intelligence. This is the first demonstration of a clinical phenotype in individual brain network topology. / 2019-07-03T00:00:00Z

Esta dissertação apresenta os principais métodos estatísticos para analisar as séries temporais de fMRI com o objetivo de detectar regiões ativadas e caracterizar o erro envolvido nessa decisão. Na análise de imagens funcionais, devido à baixa razão sinal-ruído, torna-se necessário o uso de técnicas elaboradas de processamento. O resultado da aplicação de técnicas estatísticas sobre as séries temporais obtidas da imagem de fMRI, é um mapa estatístico paramétrico, (Statistical Parametric Map), (SPM), uma imagem 3-D que permite determinar o estado do voxel, ativado ou não ativado, e a significância estatística do resultado. Propomos um novo método baseado na Difusão Anisotrópica Robusta, (Robust Anisotropic Diffusion), (RAD), que explora uma característica fundamental da imagem funcional, a correlação espacial das regiões ativadas do cérebro humano. O método proposto permite obter mapas estatísticos que melhoram a determinação das áreas ativadas a partir de dados fMRI ruidosos. Os novos mapas estatísticos paramétricos, baseados na correlação espacial da imagem fMRI, reduzem os erros do processo de classificação dos voxels, melhorando assim o mapeamento das regiões ativadas no cérebro. Aplicamos a técnica proposta em dados gerados artificialmente, simulando ruído e sinal, e avaliamos o novo método proposto e um método clássico de processamento de fMRI. Apresentamos resultados comparativos entre um método clássico, o método de correlação e o novo método. Calculamos os erros envolvidos e apresentamos a curvas características de operação de um receptor, (Receiver Operating Characteristics), (ROC), para ambos métodos, comparando os parâmetros mais importantes. Também avaliamos o novo método em dados reais de fMRI de um experimento em blocos com estímulo visual. / This dissertation presents the main statistical methods to analyse fMRI temporal series to detect activated regions and to characterise the error involved in this decision. Due to low signal to noise ratio, elaborate processing techniques are necessary to analyse functional images. Statistical techniques are usually applied on the temporal series obtained from fMRI, resulting in a Statistical Parametric Map (SPM), a 3-D image that makes it possible to determine the state of a voxel, as activated or not activated, and the statistical significance of the result. We proposed a new, simple an elegant method based on Robust Anisotropic Diffusion (RAD) to exploit the spatial correlation of activated regions of the human brain. The new method, named Robust Anisotropic Diffusion of Statistical Parametric Maps (RADSPM), allows one to obtain statistical maps that improve the determination of activated areas from noisy fMRI data. The new parametric statistical maps, based on the voxel spatial correlation of the fMRI image, reduce the classification error thus improving the significance of the results. We have tested the new technique in both simulated and real fMRI, obtaining sharp and noiseless SPMs with increased statistical significance. We compare results of the new RADSPM method with those of a classic method, the conventional correlation method. We calculate the involved errors and we present Receiver Operating Characteristics (ROC) curves for both methods, comparing the most important parameters for simulated fMRI data. We also evaluate the new method on real data of a visual fMRI experiment.

L'une des capacités humaines fondamentales est la capacité d'interpréter des symboles. Malgré plusieurs décennies de travaux en neuropsychologique et neuroimagerie sur le substrat cognitif et neuronal des représentations sémantiques, de nombreuses questions restent sans réponse. Les présents travaux de thèse tentent de démêler l'un de ces mystères: les substrats neuronaux des différentes composantes du mot sont-ils dissociables? Ce travail comporte deux composantes principales : l'une théorique et l'autre empirique. Dans la première partie, nous passons en revue les différentes positions théoriques concernant les corrélats cognitifs et neuraux des représentations sémantiques. De plus, nous proposons une distinction opérationnelle entre les dimensions moto-perceptives (c'est-à-dire les attributs des objets auxquels les mots se réfèrent perçus par les sens) et conceptuelles (c'est-à-dire l'information construite par l'intégration des multiples caractéristiques perceptives). Dans la deuxième partie, nous présentons les résultats des études menées afin d'étudier l'automaticité de la récupération, l'organisation topographique et la dynamique temporelle des dimensions moto-perceptives et conceptuelles de la signification des mots. Tout en contribuant à notre compréhension de la manière dont le sens des mots est codé dans le cerveau, les travaux présentés dans cette thèse ont des implications méthodologiques et théoriques importantes. En particulier, ils soulignent l'importance d'une intégration fructueuse entre les théories cognitives et les méthodes statistiques avancées afin d'éclairer les mystères entourant les représentations sémantiques. / One of the core human abilities is that of interpreting symbols. Notwithstanding decades of neuropsychological and neuroimaging work on the cognitive and neural substrate of semantic representations, many questions are left unanswered. The research in this dissertation attempts to unravel one of them: are the neural substrates of different components of concrete word meaning dissociated? In the first part, I review the different theoretical positions and empirical findings on the cognitive and neural correlates of semantic representations. Crucially, I propose an operational distinction between motor-perceptual dimensions (i.e., those attributes of the objects referred to by the words that are perceived through the senses) and conceptual ones (i.e., the information that is built via a complex integration of multiple perceptual features). In the second part, I present the results of the studies I conducted in order to investigate the automaticity of retrieval, topographical organization, and temporal dynamics of motor-perceptual and conceptual dimensions of word meaning. The results suggest that the neural substrates of different components of symbol meaning can be dissociated in terms of localization and of the feature of the signal encoding them, while sharing a similar temporal evolution.

Esta dissertação apresenta os principais métodos estatísticos para analisar as séries temporais de fMRI com o objetivo de detectar regiões ativadas e caracterizar o erro envolvido nessa decisão. Na análise de imagens funcionais, devido à baixa razão sinal-ruído, torna-se necessário o uso de técnicas elaboradas de processamento. O resultado da aplicação de técnicas estatísticas sobre as séries temporais obtidas da imagem de fMRI, é um mapa estatístico paramétrico, (Statistical Parametric Map), (SPM), uma imagem 3-D que permite determinar o estado do voxel, ativado ou não ativado, e a significância estatística do resultado. Propomos um novo método baseado na Difusão Anisotrópica Robusta, (Robust Anisotropic Diffusion), (RAD), que explora uma característica fundamental da imagem funcional, a correlação espacial das regiões ativadas do cérebro humano. O método proposto permite obter mapas estatísticos que melhoram a determinação das áreas ativadas a partir de dados fMRI ruidosos. Os novos mapas estatísticos paramétricos, baseados na correlação espacial da imagem fMRI, reduzem os erros do processo de classificação dos voxels, melhorando assim o mapeamento das regiões ativadas no cérebro. Aplicamos a técnica proposta em dados gerados artificialmente, simulando ruído e sinal, e avaliamos o novo método proposto e um método clássico de processamento de fMRI. Apresentamos resultados comparativos entre um método clássico, o método de correlação e o novo método. Calculamos os erros envolvidos e apresentamos a curvas características de operação de um receptor, (Receiver Operating Characteristics), (ROC), para ambos métodos, comparando os parâmetros mais importantes. Também avaliamos o novo método em dados reais de fMRI de um experimento em blocos com estímulo visual. / This dissertation presents the main statistical methods to analyse fMRI temporal series to detect activated regions and to characterise the error involved in this decision. Due to low signal to noise ratio, elaborate processing techniques are necessary to analyse functional images. Statistical techniques are usually applied on the temporal series obtained from fMRI, resulting in a Statistical Parametric Map (SPM), a 3-D image that makes it possible to determine the state of a voxel, as activated or not activated, and the statistical significance of the result. We proposed a new, simple an elegant method based on Robust Anisotropic Diffusion (RAD) to exploit the spatial correlation of activated regions of the human brain. The new method, named Robust Anisotropic Diffusion of Statistical Parametric Maps (RADSPM), allows one to obtain statistical maps that improve the determination of activated areas from noisy fMRI data. The new parametric statistical maps, based on the voxel spatial correlation of the fMRI image, reduce the classification error thus improving the significance of the results. We have tested the new technique in both simulated and real fMRI, obtaining sharp and noiseless SPMs with increased statistical significance. We compare results of the new RADSPM method with those of a classic method, the conventional correlation method. We calculate the involved errors and we present Receiver Operating Characteristics (ROC) curves for both methods, comparing the most important parameters for simulated fMRI data. We also evaluate the new method on real data of a visual fMRI experiment.

O objetivo deste trabalho é apresentar os métodos originais em Bioinformática desenvolvidos para a análise estatística na interpretação dos dados de duas técnicas baseadas em imagens: a técnica de microarranjos de DNA e a técnica de ressonância magnética funcional. O interesse principal é abordar essas técnicas experimentais quando enfrenta-se uma situação clara de amostras escassas, isto é, quando existem relativamente poucas observações experimentais do fenômeno estudado, sendo a análise individual/personalizada o representante extremo desta situação, que tem que ser resolvida. Para tanto, opta-se pelo uso da Inferência Bayesiana no contexto da Teoria da Decisão sob Incerteza, implementada computacionalmente sob o arcabouço dos Sistemas de Suporte à Decisão. Ambas as tecnologias estudadas produzem dados complexos, baseados na interpretação das diferenças entre imagens obtidas da resposta do sistema a um estímulo e da resposta numa situação controle. O resultado deste trabalho é o desenvolvimento de dois sistemas de suporte à decisão, chamados HTself e Dotslashen, para a análise de dados de microarranjos e ressonância magnética funcional, respectivamente; e de seus métodos matemáticos/computacionais subjacentes. Os sistemas desenvolvidos extraem conhecimento racional de bancos-de-dados normativos, através de modelos matemáticos específicos, contornando então o problema de amostras escassas. Finalmente, neste trabalho são descritas aplicações a problemas reais, para destacar a utilidade dos sistemas de suporte à decisão desenvolvidos nas áreas de Biologia Molecular e Neuroimagem Funcional. / The goal of this work is to present the novel Bioinformatics methods that were developed aiming the statistical analysis of two image-based techniques: DNA microarrays and functional magnetic resonance imaging. The main interest is to approach these experimental techniques in small sample size situations, i.e., when there are relatively few experimental observations of the phenomena of interest, for which the case of single subject/datum analysis is its most extreme. In order to approach these problems we chose to use Bayesian Inference in the context of the Decision Theory under Uncertainty, computationally implemented under the Decision Support Systems framework. Both technologies produce complex data, based on the interpretation of differences between images from the response to a given stimulus and the control situation. The result of this work is the development of two decision support systems, called HTself and Dotslashen, to analyze microarray and functional magnetic resonance imaging data, respectively; and the underling mathematical and computational methods. These systems use the rational knowledge from normative databases implemented in specific mathematical models, overcoming the problem of small sample size. Finally, in this work it is described applications to real problems in order to stress the utility for Molecular Biology and Functional Neuroimaging of the developed decision support systems.

The ability of the brain to change and form new neuropathways after brain injury is remarkable. The current study investigates the brains ability to form new pathways for language processing following traumatic brain injury (TBI), specifically a left temporal lobectomy. Two subjects participated in this study; one participant with TBI and one age-matched control. Sentence stimuli consisted of four types: semantically correct, semantically incorrect, syntactically correct, and syntactically incorrect. Participants underwent a fMRI scan while the auditory stimuli were presented in four blocks. Participants were asked to record if the sentence was correct or incorrect by pressing the corresponding button. It was found that reaction times for both the participant with TBI and the control were longer for the incorrect conditions. The participant with TBI generally had longer reaction times compared to the control participant and had more errors. During the fMRI scans, patient movement occurred. The block design was not set up to account for movement. Due to this factor, imaging results are questionable. While there were differences between the participant with TBI and the control participant, these differences are expected to be much larger in someone with this degree of brain injury. It is recommended for further studies to be conducted in this area with a revised block design to account for patient movement.

Remembering previously experienced events can happen as a result of an effortful retrieval attempt. At other occasions, a memory can enter our minds without any apparent effort – or, indeed, intention - to retrieve. Although it has long been appreciated that retrieval from episodic memory is intertwined with cognitive control, the neural mechanisms of memory-control interactions remain unclear. In this thesis I have used functional magnetic resonance imaging (fMRI) and scalp-recorded event-related potentials (ERP) to study the neural basis of episodic retrieval at varying levels of cognitive control. The dorsolateral prefrontal cortex (dlPFC) has been suggested to support a cognitive control mechanism (context processing) which is relevant during various situations that demand maintenance of current goals and rules. Although increased dlPFC recruitment with increasing context processing demands has been demonstrated during episodic retrieval, there are relatively few studies directly comparing the engagement of dlPFC during episodic retrieval with that during other task domains. In Study I, context processing demands were amplified in episodic retrieval, auditory attention and emotion regulation tasks. This led to overlapping dlPFC recruitment in the first two domains and a divergent reliance on ventromedial prefrontal cortex in the emotion domain. Thus, when selection between competing representations needs to be carried out in accordance with the currently relevant goals and task rules, the episodic memory system interacts with domain-general cognitive control mechanisms. Studies II and III explored the reactive nature of retrieval-specific control mechanisms: can we flexibly switch between semantic and episodic retrieval based on the information extracted from a retrieval cue? This was studied using a recognition memory task where the relevant information could with equal probability be supplied by the semantic or the episodic memory system. The fMRI results (Study II) showed that the brain activation during the ‘episodic’ but not the ‘semantic’ trials was expressed in the right prefrontal cortex. As the order of trials was unpredictable, the corresponding changes in brain activation might be evoked by differences in early cue-trace interactions. An event-related potential study (Study III) with the same experimental protocol as in Study II showed that neural processing corresponding to the two trial types diverged as early as in the time window 100-140 ms post-cue onset, thus highlighting the importance of early cue-trace matching in the selection of further retrieval processing. Study IV explored incidental episodic retrieval. Although this form of retrieval is a common experience in everyday life and a disturbing symptom in some psychiatric conditions, it is not clear how such spontaneous expressions of memory are initiated and to what extent the prefrontal cortex is engaged. The fMRI results showed, consistent with Study I, that dlPFC is specifically associated with the intention to retrieve, independently of success. Retrieval success engaged similar networks for incidentally as well as intentionally retrieved memories, comprising the hippocampus, precuneus, ventrolateral PFC, and the anterior cingulate cortex. Collectively, the fMRI and ERP results indicated that incidental retrieval was initiated by early (&lt; 200 ms) oldness estimation carried out on the semantic information extracted from the retrieval cues. Taken together, the results of this thesis indicate that episodic retrieval can be initiated via two routes:  a bottom-up input rising early during the cue processing, and a top-down input provided by the cognitive control processes mediated by the prefrontal cortex.

Functional Magnetic Resonance Imaging (fMRI) is one of the best techniques for neuroimaging and has revolutionized the way to understand the brain functions. It measures the changes in the blood oxygen level-dependent (BOLD) signal which is related to the neuronal activity. Complexity of the data, presence of different types of noises and the massive amount of data makes the fMRI data analysis a challenging one. It demands efficient signal processing and statistical analysis methods.  The inference of the analysis is used by the physicians, neurologists and researchers for better understanding of the brain functions.      The purpose of this study is to design a toolbox for fMRI data analysis. It includes methods to detect the brain activity maps, estimation of the hemodynamic response (HDR) and the connectivity of the brain structures. This toolbox provides methods for detection of activated brain regions measured with Bayesian estimator. Results are compared with the conventional methods such as t-test, ordinary least squares (OLS) and weighted least squares (WLS). Brain activation and HDR are estimated with linear adaptive model and nonlinear method based on radial basis function (RBF) neural network. Nonlinear autoregressive with exogenous inputs (NARX) neural network is developed to model the dynamics of the fMRI data.  This toolbox also provides methods to brain connectivity such as functional connectivity and effective connectivity.  These methods are examined on simulated and real fMRI datasets.