Abstract In this dissertation, we aim to study the acupuncture effect on stroke rehabilitation. To achieve this purpose, a monitoring system for performance evaluation of motor task in functional MRI environment is developed. The hypothesis of the clinical study is that acupuncture stimulation at motor zone of scalp acupoint could modulate the brain activation of motor system. Such modulation effect can be explored by fMRI. So we use the monitoring system can be used to observe patients with predominantly motor deficit without remarkable speech problem. Besides the fMRI findings of motor activation and clinical motor performance will be compared. The monitoring system must be magnetic field compatible, and furthermore, it can also detect very subtle motor performance in the case of stroke during recovery. So the specific requirements of the system challenge the need. The system includes two subsystems. One collects the position signal using a data glove and the other acquires the force signal with a pneumatic system. The understanding of the correlation of the motor task and the brain activation in depth through the integration of the position and force data can therefore be greatly improved. The monitoring system will extend to be the estimation of recovery through the integration of motor task and fMRI. Keyword: functional MRI¡Bmonitoring system¡Bmotor task¡Bforce measurement

BOLD fMRI (Blood-Oxygenation Level Dependent functional Magnetic Resonance Imaging) measures the haemodynamic correlates of brain function, with research and clinical applications. However, fMRI is limited by relatively weak signal, and large, complex noise sources. A variety of preprocessing algorithms have been developed to remove artifacts and improve signal detection, but there is no literature consensus on optimal preprocessing strategies. Furthermore, it is not well understood how fMRI experimental design choices interact with preprocessing steps. This thesis develops a statistical framework for selecting the set of preprocessing choices (“pipelines”), using data-driven metrics of (R) reproducibility of brain maps, and (P) prediction of experimental stimuli. These metrics were used to evaluate standard pipeline steps on data from young healthy subjects, who performed a set of brief tasks in an fMRI cognitive assessment battery. It is shown that (1) preprocessing choices have significant, consistent effects on the detection of brain networks in fMRI. However, (2) optimizing pipelines on a subject- and task-specific basis, compared to the standard fMRI approach of applying a single fixed set of preprocessing choices, improves (P, R) and independent test measures of between-subject activation overlap. This indicates that signal detection in standard fMRI may be limited by sub-optimal pipeline choices. Even after optimizing standard pipeline choices, physiological noise is a major confound in fMRI analysis; this includes BOLD signal changes due to respiration and pulsatile blood flow. As a potential solution, the PHYCAA (PHYsiological correction using Canonical Autocorrelation Analysis) algorithm is developed. This multivariate, data-driven model estimates physiological noise, without respiratory and cardiac measurements. The estimated noise has a spatial distribution consistent with non-neuronal tissues, and its dimensionality is correlated with cardiac and respiratory variability. Removing this physiological noise increases (P, R) of analysis results. The PHYCAA model provides novel information about the structure of physiological noise in fMRI, and a principled method of removing physiological artifact. The results of this thesis were obtained using data from a prototype fMRI cognitive assessment battery, designed for clinical use. The datasets involve brief scanning sessions with complex cognitive tasks. These findings are therefore relevant for clinical implementation of fMRI.

BOLD fMRI (Blood-Oxygenation Level Dependent functional Magnetic Resonance Imaging) measures the haemodynamic correlates of brain function, with research and clinical applications. However, fMRI is limited by relatively weak signal, and large, complex noise sources. A variety of preprocessing algorithms have been developed to remove artifacts and improve signal detection, but there is no literature consensus on optimal preprocessing strategies. Furthermore, it is not well understood how fMRI experimental design choices interact with preprocessing steps. This thesis develops a statistical framework for selecting the set of preprocessing choices (“pipelines”), using data-driven metrics of (R) reproducibility of brain maps, and (P) prediction of experimental stimuli. These metrics were used to evaluate standard pipeline steps on data from young healthy subjects, who performed a set of brief tasks in an fMRI cognitive assessment battery. It is shown that (1) preprocessing choices have significant, consistent effects on the detection of brain networks in fMRI. However, (2) optimizing pipelines on a subject- and task-specific basis, compared to the standard fMRI approach of applying a single fixed set of preprocessing choices, improves (P, R) and independent test measures of between-subject activation overlap. This indicates that signal detection in standard fMRI may be limited by sub-optimal pipeline choices. Even after optimizing standard pipeline choices, physiological noise is a major confound in fMRI analysis; this includes BOLD signal changes due to respiration and pulsatile blood flow. As a potential solution, the PHYCAA (PHYsiological correction using Canonical Autocorrelation Analysis) algorithm is developed. This multivariate, data-driven model estimates physiological noise, without respiratory and cardiac measurements. The estimated noise has a spatial distribution consistent with non-neuronal tissues, and its dimensionality is correlated with cardiac and respiratory variability. Removing this physiological noise increases (P, R) of analysis results. The PHYCAA model provides novel information about the structure of physiological noise in fMRI, and a principled method of removing physiological artifact. The results of this thesis were obtained using data from a prototype fMRI cognitive assessment battery, designed for clinical use. The datasets involve brief scanning sessions with complex cognitive tasks. These findings are therefore relevant for clinical implementation of fMRI.

Les mécanismes précis à l’origine de la dépression ne sont pas encore élucidés. L’avènement des techniques de neuro-imagerie fonctionnelle telle que l’imagerie par résonance magnétique fonctionnelle (fMRI) fournit un outil puissant permettant non seulement de définir les circuits neurobiologiques perturbés dans la dépression, mais aussi de mieux comprendre la contribution de chaque région. Les objectifs de ce travail ont consisté à mettre en évidence : (i) les clusters neuronaux impliqués dans l’évaluation hédonique d’un odorat, (ii) les anomalies cérébrales fonctionnelles sous-tendant les déficits olfactifs dans l’épisode de dépression caractérisée (EDC) et (iii) les modulations de ces anomalies suite à un traitement d’antidépresseur. Trente-huit patients dépressifs et trente sujets sains ont été sélectionnés pour réaliser un examen de fMRI comprenant trois tâches olfactives, les trois odeurs sont la menthe crépue (odeur agréable), bois de santal (odeur neutre) et le lie de vin (odeur désagréable). D’après notre étude nous avons conclu que les patients déprimés présentent des anomalies de fonctionnement dans le thalamus qui peut être considéré un marqueur efficace pour le pronostic de la dépression. De plus, la fMRI constitue un bon outil pour juger de l’efficacité du traitement antidépresseur. / The precise mechanisms at the origin of the depression are not yet elucidated. The advent of functional neuroimaging techniques such as functional magnetic resonance imaging (fMRI) has provided a powerful tool not only for defining the neurobiological circuits disturbed in depression, but also for better understanding the contribution of each region. The objectives of this work were to highlight: (i) the clusters involved in the hedonic evaluation of the smell (ii) the functional brain abnormalities underlying the olfactory deficits in the major depressive episode (MDE) and (iii) the modulations of these abnormalities following antidepressant treatment. Thirty-eight depressive patients and thirty healthy subjects were selected to perform a fMRI examination with three tasks: recognizing three smells which are the spearmint, sandalwood and wine lees. Based on our study, we can conclude that depressed patients have functional abnormalities in the thalamus, this region may be considered as a good marker for the prognosis of depression. In addition, fMRI could be a good tool to evaluate the treatment performance of antidepressant treatment.

No description available.

No description available.

Clinical and behavioural evidence has shown that the threat value of pain biases decisions about whether a stimulus is perceived as painful or not, and if yes, how intense is the sensation. This thesis aims to investigate the neural mechanisms underlying the effect of perceived threat on perceptual decisions about pain. The first study investigates the neural mechanisms underlying the effect of threat on the decision about the quality of the sensation, i.e., whether it is perceived as painful or not. The perception of pain (relative to no pain) was associated with activation in the anterior insula as well as an increased connectivity between this region and the mid-cingulate cortex (MCC). Activity in the MCC was correlated with the threat-related bias to perceived pain. In the second study, probabilistic tractography was performed with diffusion tensor imaging to investigate the structural connectivity between subdivisions of the insula and other pain-related regions. Additional analyses revealed that the structural connectivity between the anterior insula and the MCC, and between the posterior insula and somatosensory cortices, is positively correlated with the threat-related bias toward pain. In the third study, a multivariate pattern analysis (MVPA) was performed to investigate whether pain can be decoded from functional neuroimaging data acquired during the anticipation and during the receipt of pain. The results show that pain can be predicted by the pattern of neural activity in the right anterior insula during anticipation and stimulation. The fourth study investigated the effect of uncertainty about the stimulation intensity as a form of threat on the perceived intensity of pain. Uncertainty was found to be associated with an increased activation in the anterior insula. Overall, these findings suggest that a neural network consisting of the anterior insula and the MCC plays a key role in decisions about the quality and the quantity of nociceptive sensation. Results from the MVPA analysis support the notion that perceptual decisions are encoded by a distributed network of brain regions. The variability in anatomical connections between these regions may account for the individual differences in the susceptibility to a threat-mediated bias toward pain.

Chronic pain is a major health problem that affects approximately 20% of the adult population, but only 60% of its patients find success in managing their condition. As an alternative therapeutic tool, transcranial direct current stimulation (tDCS) has appeared promising in recent literature, with several papers using the left dorsolateral prefrontal cortex (L-DLPFC) as the stimulation target due to its inherent role in pain modulation. However, the underlying mechanisms of this treatment have never been directly investigated. The thesis intends to explore this question through the application of a combination of tDCS, functional and structural imaging, and an ongoing pain model. The first study investigated the task-free effects of L-DLPFC tDCS by using concurrent arterial spin labeling (ASL) perfusion imaging. Anodal (excitatory) tDCS was associated with an increase in regional cerebral blood flow (rCBF) of L-DLPFC, while cathodal (inhibitory) tDCS was related to a relative decrease. Regions connected to L-DLPFC, such as thalamus and primary somatosensory cortex, also experienced significant perfusion changes. Further analyses found modulations of L-DLPFC–thalamic functional correlations, which was particularly relevant due to the importance of the thalamus in antinociceptive processing. To understand pain neuromodulation, a robust ongoing pain model was required to serve as a platform for the investigation. Thus, electrical, thermal, and capsaicin stimuli were tested in a series of studies. The former two did not produce an ongoing pain sensation, thus failing to replicate a chronic pain experience on healthy volunteers. However, topical application of capsaicin appeared to induce such a response, which entailed a pain intensity rating between 5 and 7 on a 0-10 visual analogue scale for at least 30 minutes. Careful screening ensued to select capsaicin responders from the cohort, who served as the research volunteers for the remainder of the thesis. Combining these two studies, rCBF changes associated with tDCS-induced pain neuromodulation were then directly investigated with concurrent ASL and the above capsaicin model. Behavioral measurements suggested a significant reduction in pain intensity for the anodal tDCS condition over the sham tDCS condition, which corroborated the results from previous pain-tDCS studies. Modulations of L-DLPFC and thalamic rCBF continued to be observed with anodal tDCS. In addition, the activity of the posterior insula, an indicator of ongoing pain intensity, was higher for the sham tDCS condition than for the anodal tDCS condition. The final study explored the importance of L-DLPFC–thalamic connection in pain modulation with diffusion tensor imaging (DTI). Probability tractography and tract-based spatial statistics (TBSS) were used to extract the mean fractional anisotropy (FA) of the tract. As hypothesized, the mean FA values within this tract positively correlated with the pain intensity changes, which indicated that individuals with stronger structural connection enjoyed greater alleviation from L-DLPFC tDCS. As a whole, the above studies suggest that the clinical efficacy of L-DLPFC tDCS in pain treatment may arise from the resulting top-down modulation of the thalamus. Further studies on chronic pain patients may offer further verification of the mechanism that has been proposed in this thesis.

La conception classique de l'organisation des réseaux cérébraux participant au langage décrit deux zones corticales, l'une frontale (Broca), l'autre temporale intervenant respectivement dans la production et la compréhension du langage, unies par un faisceau de substance blanche: le faisceau longitudinal supérieur. L'imagerie par résonance magnétique d'activation (IRMf) a rendu possible la visualisation de zones d'activation corticales, et l'imagerie en tenseur de diffusion avec la tractographie celle des faisceaux de substance blanche. Ces données nouvelles permettent de repenser l'organisation corticale et sous-corticale du langage. L'ensemble des travaux en imagerie d'activation étaye l'hypothèse d'une dissociation dorso-ventrale du traitement du langage. Les zones de traitement de l'information phonétique étant dorsales, et les centres impliqués dans le traitement sémantique plus ventraux. Imagerie d'activation, imagerie du tenseur de diffusion, stimulations corticales et sous-corticales ont permis d'aboutir à un modèle de réseau du langage impliquant une voie dorsale, essentiellement phonologique et une voie ventrale, sémantique présentant deux composantes. La première est directe, par le faisceau fronto-occipital inférieur connectant les aires temporales postérieures à la région orbito-frontale. La seconde est une voie indirecte qui connecte successivement la région occipito-temporale au pôle temporal par l'intermédiaire du faisceau longitudinal inférieur puis le pôle temporal aux aires basifrontales par le faisceau unciné. Cette dernière voie est compensée après résection ou lors des stimulations per-opératoires, suggérant la possibilité de réseaux de suppléance parallèles et bilatéraux. Dans cette vision d'une voie ventrale sémantique bilatérale, notre travail a pour objet par l'utilisation de l'imagerie fonctionnelle d'activation, d'une part, et de l'imagerie en tenseur de diffusion d'autre part, 1) de vérifier le nombre de faisceaux qui sous-tendent cette voie et en particulier de la part attribuable au faisceau longitudinal moyen; 2) de caractériser les réorganisations de la connectivité fonctionnelle du réseau du langage après une chirurgie de tumeur gliale hémisphérique gauche. Dans la première partie de ce travail, après avoir rappelé la filiation phylogénétique de l'homme et du primate non-humain, en insistant sur les dissemblances qui existent dans leur anatomie corticale et sous-corticale, nous exposons comment ces données ont permis d'aboutir à la découverte du faisceau longitudinal moyen chez l'homme. Nous confirmons la visibilité en imagerie du tenseur de diffusion de ce faisceau et nous précisons ses rapports avec les différents faisceaux de substance blanche constitutifs des voies ventrales et dorsales du langage. Enfin nous discutons ces données tractographiques à la lumière de la dissection et discutons du rôle présumé du faisceau dans le langage. Dans un deuxième temps, nous présentons les concepts de connectivité anatomique, fonctionnelle et effective. Puis nous appliquons un outil d'analyse de la connectivité fonctionnelle à des données périopératoires en tâches de fluence afin de réaliser une cartographie de la plasticité périopératoire de la composante sémantique du langage. Les résultats de cette deuxième étude suggèrent un recrutement de l'hémisphère ipsilatéral à la lésion au décours de la chirurgie. Enfin, nous évoquons la possibilité d'induire une désynchronisation (i.e. une altération de la connectivité) du réseau du mode par défaut par une stimulation peropératoire de la partie postérieure du faisceau cingulaire, induisant une sensation de dépersonnalisation, ces résultats suggérant qu'une des fonctions du noeud le plus postérieur du réseau soit de maintenir un état de conscience du monde extérieur. / According to classical conception of the anatomo-functional organization of language, there are two main cortical areas: a frontal area (Broca) and a temporal one (Wernicke) respectively involved in language production and comprehension. Functional magnetic resonance Imaging (fMRI) reveals cortical areas of activation and diffusion tensor imaging-based tractography (DTI) makes feasible the visualization of white-matter tracts in the human brain. On the basis of these techniques, a new conception of language cortical and sub-cortical organization arose, supporting the hypothesis that language processing network is dissociated in an dorso-ventral way. Dorsal areas of the brain being devoted to phonologic processing and its ventral areas to semantic processing of speech. Different techniques such as fMRI, DTI, intraoperative cortical and subcortical mapping made possible to describe two pathways involved in the language network: a dorsal stream and a ventral stream respectively involved in phonologic and semantic processing. As the dorsal route is composed of a unique pathway, the superior longitudinal fasciculus, the ventral stream appears to be composed of two different pathways. First a direct pathway, the inferior occipitofrontal fasciculus, connecting the posterior temporal areas to the orbitofrontal region. The second one, an indirect pathway, the inferior longitudinal fasciculus links the posterior occipitotemporal to the temporal pole, then relayed by the uncinate fasciculus connecting the temporal pole to the basifrontal areas. According to these observations the aim of our work is In the present work, we aim (1) to confirm that the MdLF is constantly found in control subjects and that it can be delineated from the other fiber tracts that constitute language pathways, (2) to characterize the reorganization of language network's functional connectivity follmowing surgical removal of left hemisphere low grade gliomas. In this study, we confirmed that the MdLF is constantly found in healthy volunteers and we clearly delineate the MdLF from the other fascicles that constitute language pathways, especially the ventral pathway. Considering language plasticity, our findings suggest that in the postoperative period, brain plasticity occurs with an ipsilateral recrutment and increased fonctional connectivity in the left hemisphere. Finally, we report a collaborative work observing that intraoperative electrostimulations of the white matter underlying the left posterior cingulate, while performing a naming task, systematically induced an unresponsive state for few seconds in relationship with a dream-like state. This result provides direct evidence that connectivity underlying the posterior node of the default mode network permits maintained consciousness of the external world.

Activation in the hippocampus (HC) and prefrontal cortex (PFC) is critical to accurately retrieve overlapping sequences. Experiments 1 and 2 tested the hypotheses that activation in and interaction between HC and PFC increases as overlap between sequences increases in a non-spatial task. Experiment 3 tested the hypothesis that theta oscillations are involved in orchestrating interactions between HC and PFC in a spatial task with overlapping elements. In the first two studies, 17 participants (aged 18-34; 11 female) learned sequences consisting of a picture frame, face, and scene. Conditions varied by degree of overlap. Using fMRI, Experiment 1 tested how degree of overlap affected HC and PFC activation. In overlapping sequences, middle and posterior HC were active when predictability of the correct response increased, dorsolateral PFC was active when participants were able to ascertain the correct set of sequences, and ventrolateral PFC was active when inhibition of interfering associations was required. Experiment 2 examined functional connectivity of HC and PFC during disambiguation. Low- and high-overlap conditions were associated with increased connectivity in separate regions at different times indicating that retrieval under the two conditions used different neural networks and strategies. Low-overlap trials were associated with increased connectivity between HC and prefrontal and parietal regions. High-overlap trials showed increased connectivity between lateral PFC and visual areas, indicating that imagery may be necessary for accurate performance. Using EEG recording, Experiment 3 examined theta activity during retrieval of well-learned, overlapping and non-overlapping mazes in 17 participants (aged 18-34, 11 female). Theta activity increased in overlapping mazes during the first of four hallways, suggesting participants were looking ahead to upcoming turns in the maze. Theta activity increased at the beginning and choice point of the third overlapping hallway, possibly in response to interference from the paired, overlapping maze. These studies provide evidence that (1) overlapping associations in non-spatial sequences elicit interactions between hippocampus and lateral prefrontal cortex, (2) increasing the degree of overlap changes the neural processes required to perform the task, and (3) theta power increases in response to increased cognitive demand and maintenance of sequence information needed to differentiate between overlapping spatial routes.