Event-related brain potentials (ERPs) in electroencephalography are manifestations at the scalp of neural activity that is triggered by, and is involved in, the processing of specific events. This chapter focuses on braincomputer interfaces (BCIs) that use P300, an endogenous ERP component. The P300 is a positive potential that occurs over central-parietal scalp 250- 700 msec after a rare event occurs in the context of the oddball paradigm. This paradigm has three essential attributes: a subject is presented with a series of events (i.e., stimuli), each of which falls into one of two classes; the events that fall into one of the classes are less frequent than those that fall into the other class; and the subject performs a task that requires classifying each event into one of the two classes.

Le sommeil est un phénomène biologique universel complexe et encore peu compris. La méthode de référence actuelle pour caractériser les états de vigilance au cours du sommeil est la polysomnographie (PSG) qui enregistre de manière non invasive à la surface de la peau, les modifications électrophysiologiques de l’activité cérébrale (électroencéphalographie, EEG), oculaire (électro-oculographie, EOG) et musculaire (électromyographie, EMG). Traditionnellement, les signaux électrophysiologiques sont ensuite analysés par un expert du sommeil qui annote manuellement les évènements d’intérêt comme les stades de sommeil ou certains micro-évènements (grapho éléments EEG). Toutefois, l’annotation manuelle est chronophage et sujette à la subjectivité de l’expert. De plus, le développement exponentiel d’outils de monitoring du sommeil enregistrant et analysant automatiquement les signaux électrophysiologiques tels que le bandeau Dreem rend nécessaire une automatisation de ces tâches.L’apprentissage machine bénéficie d’une attention croissante car il permet d’apprendre à un ordinateur à réaliser certaines tâches de décision à partir d’un ensemble d’exemples d’apprentissage et d’obtenir des performances de prédictions plus élevées qu’avec les méthodes classiques. Les avancées techniques dans le domaine de l’apprentissage profond ont ouvert de nouvelles perspectives pour la science du sommeil tout en soulevant de nouveaux défis techniques. L’entraînement des algorithmes d’apprentissage profond nécessite une grande quantité de données annotées qui n’est pas nécessairement disponible pour les données PSG. De plus, les algorithmes d’apprentissage sont très sensibles à la variabilité des données qui est non négligeable en ce qui concerne ces données. Cela s’explique par la variabilité intra et inter-sujet (pathologies / sujets sains, âge…).Cette thèse étudie le développement d’algorithmes d’apprentissage profond afin de réaliser deux types de tâches: la prédiction des stades de sommeil et la détection de micro-événements. Une attention particulière est portée (a) sur la quantité de données annotées requise pour l’entraînement des algorithmes proposés et (b) sur la sensibilité de ces algorithmes à la variabilité des données. Des stratégies spécifiques, basées sur l’apprentissage par transfert, sont proposées pour résoudre les problèmes techniques dus au manque de données annotées et à la variabilité des données. / Sleep is a complex and not fully understood biological phenomenon. The traditional process to monitor sleep relies on the polysomnography exam (PSG). It records, in a non invasive fashion at the level of the skin, electrophysiological modifications of the brain activity (electroencephalography, EEG), ocular (electro-oculography, EOG) and muscular (electro-myography, EMG). The recorded signals are then analyzed by a sleep expert who manually annotates the events of interest such as the sleep stages or some micro-events. However, manual labeling is time-consuming and prone to the expert subjectivity. Furthermore, the development of sleep monitoring consumer wearable devices which record and process automatically electrophysiological signals, such as Dreem headband, requires to automate some labeling tasks.Machine learning (ML) has received much attention as a way to teach a computer to perform some decision tasks automatically from a set of learning examples. Furthermore, the rise of deep learning (DL) algorithms in several fields have opened new perspectives for sleep sciences. On the other hand, this is also raising new concerns related to the scarcity of labeled data that may prevent their training processes and the variability of data that may hurt their performances. Indeed, sleep data is scarce due to the labeling burden and exhibits also some intra and inter-subject variability (due to sleep disorders, aging...).This thesis has investigated and proposed ML algorithms to automate the detection of sleep related events from raw PSG time series. Through the prism of DL, it addressed two main tasks: sleep stage classification and micro-event detection. A particular attention was brought (a) to the quantity of labeled data required to train such algorithms and (b) to the generalization performances of these algorithms to new (variable) data. Specific strategies, based on transfer learning, were designed to cope with the issues related to the scarcity of labeled data and the variability of data.

京都大学 / 0048 / 新制・課程博士 / 博士(情報学) / 甲第19126号 / 情博第572号 / 新制||情||100(附属図書館) / 32077 / 京都大学大学院情報学研究科システム科学専攻 / (主査)教授 石井 信, 教授 田中 利幸, 教授 加納 学 / 学位規則第4条第1項該当 / Doctor of Informatics / Kyoto University / DFAM

京都大学 / 0048 / 新制・論文博士 / 博士(医学) / 乙第13125号 / 論医博第2134号 / 新制||医||1024(附属図書館) / (主査)教授 宮本 享, 教授 村井 俊哉, 教授 井上 治久 / 学位規則第4条第2項該当 / Doctor of Medical Science / Kyoto University / DFAM

The human brain relies on the complex interactions of distinct brain regions to support cognitive processes. The interplay between the hippocampus and neocortical regions plays a key role in the formation, storage, and retrieval of long-term episodic memories. Oscillatory activities during sleep are a fundamental mechanism that binds distributed neuronal networks into functionally coherent ensembles. However, the large-scale hippocampal-neocortical oscillatory mechanisms that support flexible modulation of long-term memory remain poorly understood. Furthermore, alterations to physiologic spatiotemporal patterns that are essential for intact memory function can result in pathophysiology in brain disorders such as focal epilepsy. Investigating how epileptic network activity disrupts connectivity in distributed networks and the organization of oscillatory activity are additional crucial areas that require further research. Our experiments on rodents and human patients with epilepsy have provided valuable insights into these mechanisms. In rodents, we used high-density microelectrode arrays in tandem with hippocampal probes to analyze intracranial electroencephalography (iEEG) from multiple cortical regions and the hippocampus. We identified key hippocampal-cortical oscillatory biomarkers that were differentially coordinated based on the age, strength, and type of memory. We also analyzed iEEG from patients with focal epilepsy and demonstrated how individualized pattern of pathologic-physiologic coupling can disrupt large-scale memory circuits. Our findings may offer new opportunities for therapies aimed at addressing distributed network dysfunction in various neuropsychiatric disorders.

Objective: To assess the reliability and usefulness of an EEG-based brain-computer interface (BCI) for patients with advanced amyotrophic lateral sclerosis (ALS) who used it independently at home for up to 18 months. Methods: Of 42 patients consented, 39 (93%) met the study criteria, and 37 (88%) were assessed for use of the Wadsworth BCI. Nine (21%) could not use the BCI. Of the other 28, 27 (men, age 28-79 years) (64%) had the BCI placed in their homes, and they and their caregivers were trained to use it. Use data were collected by Internet. Periodic visits evaluated BCI benefit and burden and quality of life. Results: Over subsequent months, 12 (29% of the original 42) left the study because of death or rapid disease progression and 6 (14%) left because of decreased interest. Fourteen (33%) completed training and used the BCI independently, mainly for communication. Technical problems were rare. Patient and caregiver ratings indicated that BCI benefit exceeded burden. Quality of life remained stable. Of those not lost to the disease, half completed the study; all but 1 patient kept the BCI for further use. Conclusion: The Wadsworth BCI home system can function reliably and usefully when operated by patients in their homes. BCIs that support communication are at present most suitable for people who are severely disabled but are otherwise in stable health. Improvements in BCI convenience and performance, including some now underway, should increase the number of people who find them useful and the extent to which they are used.

No description available.

Attention reorienting is crucial to human survival in a constantly changing environment. In order to react and respond to novel and potentially threatening stimuli in the environment, we have to first reorient our attention to the stimuli themselves. While numerous studies in the past have attempted to uncover the principles of how our brain processes new stimuli and reorients our attention, they typically employed standardized paradigms such as an oddball or a cueing paradigm that do not represent how humans actually reorient attention in the real world. This dissertation seeks to directly address this issue by investigating the brain dynamics underlying attention reorienting in an immersive and naturalistic environment. We employ a virtual reality (VR)-based target detection paradigm that closely mimics how human would reorient their attention in real-world situations. During the experiments, subjects are instructed to reorient their attention between a primary visual task (driving simulation) and a secondary visual task (target detection) while their electroencephalography (EEG), eyetracking and behavioral inputs are being recorded. Each set of experiments and subsequent data analysis methods are tailored to answer different questions based on the three specific aims of this dissertation (1) how do eye movements affect attention reorienting signals? (2) how do we integrate the information obtained from the neural and ocular signals to decode reorienting? and (3) what is the relationship between attention reorienting and the arousal system? We found that while eye movements result in greater temporal variation of neural signals associated with attention reorienting, namely the P300 signal, time-locking the event-related potentials (ERPs) to image onset or saccade intersection still results in the best overall performance in classifying target vs. distractor stimuli. Similarly to eye movements, we also found that allowing for head movements results in greater temporal variations of both the neural (P300) and pupil-linked attention reorienting signals. However, by combining the EEG, pupil dilation and dwell time signals, a multi-modal hybrid classifier we developed using the hierarchical discriminant component analysis (HDCA) was able to capture and integrate the neural and ocular attention reorienting signals with similar performance both in the condition with and without head movements. In addition, the hybrid classifier outperformed single-modality classifiers (EEG-only, pupil dilation-only and dwell time-only) in all comparisons. Lastly, we reported a close-knit relationship between pupil-linked arousal and network-level EEG dynamics underlying attention reorienting. We observed improvements in overall performance as pupil-linked arousal increased. We also observed increased oscillatory activity across multiple frequency bands in regions associated with the dorsal and ventral attention networks as pupil-linked arousal increased. Additionally, we found a decrease in functional connectivity across nodes in the salience network and the ventral attention network as pupil-linked arousal increased. The findings of this dissertation have the potential to serve as the basis for the development of the next generation of non-invasive brain-computer interfaces (BCIs) that can function in real-world environments. Furthermore, these findings may also serve to help physicians and neuroscientists better understand the neurophysiology underlying attention-related disorders including attention-deficit disorder (ADD) or attention-deficit/hyperactivity disorder (ADHD).

Hearing loss, as a result of old age, has been linked to a decline in speech perception despite the use of additional listening devices. Even though the relationship between hearing loss and decreased speech perception has been well established, research in this area has often focused on the behavioral aspects of language and not on the functionality of the brain itself. In the present study, the mismatch negativity, an event related potential, was examined in order to determine the differences in speech perception between young adult participants, geriatric normal hearing participants, and geriatric hearing-impaired participants. It was hypothesized that a significantly weaker mismatch negativity would occur in the geriatric hearing-impaired participants when compared to the young adult participants and the geriatric normal hearing participants. A passive same/different discrimination task was administered to 10 young adult controls (5 male, 5 female) and eight older adult participants with and without hearing loss (4 male, 4 female). Data from behavioral responses and event related potentials were recorded from 64 electrodes placed across the scalp. Results demonstrated that the mismatch negativity occurred at various amplitudes across all participants tested; however, an increased latency in the presence of the mismatch negativity was noted for the geriatric normal hearing and the geriatric hearing-impaired participants. Dipoles reconstructed from temporal event related potential data were located in the cortical areas known to be instrumental in auditory and language processing for the young adult participants; however, within the geriatric normal hearing and the geriatric hearing-impaired participants, dipoles were seen in multiple locations not directly associated with language and auditory processing. Although not conclusive, it appears that within the geriatric normal hearing and the geriatric hearing-impaired participants there is slower processing of the speech information, as well as some cognitive confusion which leads to fewer available resources for interpretation.

Inner speech, or self-talk, is a process by which we talk to ourselves to think through problems or plan actions. Although inner speech is ubiquitous, its neural basis remains largely unknown. This thesis investigates the feasibility of using brain signal analysis to classify the recorded electroencephalography (EEG) data from participants engaged in tasks involving Inner Speech and made publicly available by Nieto et al. (2021). We present the implementation of four machine learning models, demonstrate the results, and compare using different protocols. The results are compared to the ones obtained by Berg et al. (2021), who used the same dataset. Two of the classical models we tried (SVC and LinearSVC) prove superior even against results obtained with deep learning models. We also compare the results from Inner Speech with Pronounced Speech to validate the reusability of the proposed method. We found an apparent regularity in the data on the results, validating the method’s quality and reusability.