Author:

Keywords:

EEG-fMRI, data-driven methods, cognitive processes, SISTA

Abstract:

Understanding the cognitive processes that are going on in the human brain, requires the combination of several types of observations. For this reason, since several years, neuroscience research started to focus on multimodal approaches. One such multimodal approach is the combination of electroencephalography (EEG) and functional magnetic resonance imaging (fMRI). The non-invasive character of these two modalities makes their combination not only harmless and painless, but also especially suited for widespread research in both clinical and experimental applications. Moreover, the complementarity between the high temporal resolution of the EEG and the high spatial resolution of the fMRI, allows obtaining a more complete picture of the processes under study.However, the combination of EEG and fMRI is challenging, not only on the level of the data acquisition, but also when it comes to extracting the activity of interest and interpreting the results from integrated analyses. This thesis, therefore, describes the setup of a typical EEG-fMRI study in detail and addresses the different steps to be taken in a full EEG-fMRI analysis. More specifically, a number of data-driven approaches that can be used at several stages of an EEG-fMRI study are optimized and validated. In this context, the work presented in this thesis can be subdivided in three main parts.First, preparatory to any further analysis, the EEG and fMRI data need to be preprocessed and artifacts caused by the simultaneous acquisition need to be removed from the EEG data. The removal of these artifacts, and especially of the ballistocardiogram (BCG) artifact, is not straightforward. Despite considerable effort on this issue, no consensus has yet been reached on the best removal method. Therefore, this thesis investigates some of the most widely used methods for BCG artifact removal. More in detail, this work specifically focuses on the method parameters and on an accurate validation based on different task conditions and single trial event-related potentials (ERPs).Further, the use of parallel factor analysis (PARAFAC) for the extraction of task-related characteristics from ERP data, is shown. It is demonstrated that PARAFAC can distinguish between different task-related conditions on a single trial level, by employing the PARAFAC trial signatures for classification. Also, an evaluation of the robustness of the method against noise is presented, by including data measured inside the scanner. Although the obtained accuracies are lower in the latter case, PARAFAC proves to perform better than a classification based on raw data single trial characteristics.Finally, two different approaches for EEG-fMRI integration (and more specifically, the integration of fMRI with the ERP data) are assessed. First, the single trial ERP information obtained with PARAFAC is used for the analysis of fMRI data. Second, average ERPs and fMRI from different subjects are combined in a so-called JointICA analysis. These two obviously distinct methods in fact address two different advantages of combined EEG-fMRI studies. Whereas the single trial analysis allows interpreting the possible connection between fluctuations in EEG and fMRI on a single trial level, JointICA enables the generation of a full spatiotemporal picture of the ongoing processes.As such, this thesis confirms the usefulness of data-driven methods in the analysis and integration of EEG and fMRI, thereby extending results of earlier studies in the same research field.