Children with multiple sclerosis (MS) frequently have reduced visual processing speed, which can affect learning, social interaction and daily functioning. Identification of cognitive impairment at an early stage is essential for timely intervention. This study explores if electroencephalography (EEG)-based functional connectivity (FC) can be utilized as a non-invasive biomarker for visual processing speed in children with MS.

Resting state EEG data and neuropsychological assessments were analyzed from ten children diagnosed with MS. A custom preprocessing pipeline was developed and validated to ensure data quality for FC analysis. Connectivity metrics, including fronto-occipital and interhemispheric connectivity, network efficiency based on minimum spanning tree analysis, and individual alpha peak frequency, were extracted from source-reconstructed EEG. Correlations were tested between these metrics and visual processing speed, measured with the Processing Speed Index of the Wechsler Intelligence Scale for Children.

As hypothesized, trends toward greater network efficiency and stronger interhemispheric and fronto-occipital connectivity were associated with higher visual processing speed. Although none of these associations reached statistical significance, the results support the potential of EEG-based functional connectivity to evolve into a clinically relevant biomarker, enabling early diagnosis, personalized treatment and improved prognosis for children with MS.

Code repository - https://github.com/Ilse2001/code-repository-msc-thesis-eeg-fc

Accurate early prognostication after pediatric cardiac arrest is clinically crucial yet remains methodologically challenging. This thesis frames the problem as a representation learning task on raw electroencephalography (EEG) and introduces a fully self-supervised pipeline followed by a supervised classifier. A Time-Series-to-Vector (TS2Vec) encoder is trained on 6,669 unlabeled twenty-second EEG epochs (T = 2000 time points, C = 4 channels) from 84 comatose children treated at Erasmus MC Sophia Children’s Hospital. The encoder learns to map each EEG epoch to a 320-dimensional vector by contrasting transformed views of the same signal from unrelated signals, without using outcome labels.

Downstream classification is performed with a k-nearest neighbors (k-NN) model. Probabilities are assigned at the epoch level and aggregated across all epochs of a patient to generate a final patientlevel prognosis. Five-fold patient-level cross-validation gives an area under the ROC curve of 0.861 ± 0.092, an accuracy of 0.774±0.045 and an F1 score of 0.733±0.057. Critically, specificity and precision are both 1.000 ± 0.000. Thus the model never predicts death for a child who survives, satisfying a stringent clinical safety requirement. Sensitivity is 0.581 ± 0.070, reflecting a deliberately conservative decision threshold.

Qualitative analyses offers tentative support for the physiological relevance of the learned features. t-SNE projections show clustering by patient identity and good separation between the two extreme EEG background patterns (continuous normal activity and electrocerebral silence) without supervision. Saliency analysis did not yield clinically interpretable patterns.

Compared to the leading feature-engineered qEEG baseline, the proposed approach achieves equivalent specificity and a slightly lower AUC, while removing the need for manual feature design and expert annotation. The modular architecture invites several extensions, including graph-based encoders that respect electrode topology, integration of auxiliary data such as ECG and clinical metadata, retrieval-based reasoning using large historical EEG archives, and more interpretable or end-to-end trainable models. Overall, this work demonstrates that contrastive learning can be effectively applied to raw pediatric EEG for conservative, label-efficient outcome prediction after cardiac arrest, and offers a mathematically grounded proof of concept for future clinical applications.

Introduction
Upper limb tremor is a debilitating symptom of chronic inflammatory demyelinating polyneuropathy (CIDP). It generally does not respond well to therapy, and its pathophysiology is debated. Both a peripheral and central cause of tremor in CIDP have been hypothesized, and also an interplay between the two has been proposed. In case a central pathophysiology underlies tremor in CIDP patients, the cerebello-thalamo-cortical circuit may be involved. This has not been researched in CIDP before, but this circuit is known to play a role in other tremor disorders. Functional Magnetic Resonance Imaging (fMRI) can non-invasively assess brain activity, and will be used in this thesis to investigate a potential central pathophysiology of tremor in CIDP patients.

Aim
The aim of this thesis is to assess brain activity in the cerebello-thalamo-cortical circuit in CIDP patients who suffer from tremor, and compare this to the brain activity in this circuit in CIDP patients without tremor.

Methods
fMRI scans were acquired from both CIDP patients who suffered from tremor (tremor group), and CIDP patients without tremor (control group). All participants performed two motor tasks, alternated with rest. In the static motor task, they extended their arms in front of them, inducing a postural tremor in the patients from the tremor group. In the dynamic motor task, all participants extended their arms and additionally performed a tremorlike movement with their hands. In the tremor group, an intentional tremor cooccurred with the voluntary movement of the dynamic motor task. A general linear model was used to analyse the fMRI data. Within-group and between-group analyses of clusters of brain activity were performed using one-sample and two-sample t-tests respectively.

Results
9 tremulous CIDP patients and 12 controls were included in the analysis. In the between-group analysis of the static task, more activity in CIDP patients with tremor in comparison to CIDP patients without tremor was found in lobule VIII of the cerebellum, the premotor cortex and supplementary motor areas, the visual cortex, and the fusiform. In the dynamic task, CIDP patients with tremor showed more activity in lobules IV/V of the cerebellum, the premotor cortex and supplementary motor areas, the supramarginal gyrus, the pars orbitalis, the insula, and the Rolandic operculum.

Conclusion
The results suggest that there is more brain activity in the cerebello-thalamo-cortical circuit in CIDP patients who suffer from tremor in comparison to CIDP patients without tremor. However, the results of the dynamic task may be influenced by a difference in brain activity due to the voluntary tremorlike movement. Further research is necessary to first of all separate the brain activity related to tremor from activity resulting from the voluntary movement, and furthermore to give insight into if this hyperactivity in the cerebello-thalamo-cortical circuit in CIDP patients with tremor is the cause of tremor, maintaining it, or is a reflection of the tremor.

Automated diagnosis of epilepsy for differentiating epileptic EEGs without Interictal Epileptic Discharges (IEDs) from normal EEGs remains a critical challenge in clinical settings.
Current state-of-the-art methods use algorithms that can effectively detect epilepsy seizures which improves the current treatment methods for people suffer from epilepsy. Electroencephalograms (EEGs) analyzed by neurologists which are not able to meet the criteria are further looked into to obtain an efficient classification. This thesis presents an automated epilepsy diagnosis approach using a multi-algorithmic feature extraction pipeline. The final models include the development of a robust multi-processing feature extraction pipeline, the application of advanced machine learning / deep learning algorithms, and the validation of the proposed methods on comprehensive EEG datasets. The results, achieved using an XGBoost classifier with leave-one-subject-out (LOSO) cross-validation, demonstrate comparable performance to state-of-the-art epilepsy detectors. The study emphasizes the detection of epilepsy without IEDs, optimizing models through nested cross-validation, and evaluates their performance on the Temple University Hospital (TUH) and Erasmus Medical Center (EMC) Rotterdam datasets.

The brain is one of the most important organs of the human body, yet it remains poorly understood. The anatomical connections between different brain regions facilitate the transmission of electrical signals across the brain, forming the basis of higher functionalities such as speech, vision, comprehension, and sensory perception. During brain tumor surgery, it is critical to preserve vital connections within the brain while resecting as much of the tumor as possible. Therefore, accurately locating and mapping functional tracts within the brain is crucial for optimal patient outcomes. This thesis focuses on the exploration and application of electrocorticography (ECoG) and Cortico-Cortical Evoked Potentials (CCEP) to enhance our understanding of brain connectivity to ultimately aid in the improvement of tumor resection.

The Background chapter provides all necessary information on the electrophysiological brain, intraoperative neuromonitoring, ECoG, and CCEP. By explaining the fundamental concepts and technological aspects, the chapter sets the stage for understanding how ECoG and CCEP can be utilized to study brain connectivity and improve surgical outcomes.

Chapter 1 establishes the foundation of the research. It uses data from a study on brain transmission speeds involving 74 patients who underwent epilepsy surgery. The chapter explains the composition and structure of the raw data, explaining the processes of data loading and preprocessing, including the construction of epochs and evoked objects from the raw data. The peak detection algorithm, the primary analysis method used to extract amplitude and latency parameters from registered CCEP N1 peaks, is illustrated. These results, a combination of the CCEP connections between electrodes and their corresponding amplitude and latency, form the basis for further chapters.

In Chapter 2, the research primarily addresses the analysis of CCEP responses measured with ECoG across a population. The analysis made a construction of brain connectivity graphs based on Destrieux labels and individual electrodes. The peak detection algorithm identified various CCEP responses and highlighted connections between stimulating and recording electrodes. Notably, connections between temporal and parietal brain regions, and between parietal and frontal regions, were found. These findings corresponded with known white matter tracts, demonstrating the capability of CCEP measurements to reflect connectivity, comparable to standard DTI measurements.

Chapter 3 shifts focus to the visualization of these CCEP responses at an individual patient level for clinical translation. Stimulation of electrodes in the parietal region revealed both local responses and distant connections in the frontal and temporal regions of an exemplary patient. These visualizations provide intuitive insights into the brain's connectivity. An animation illustrating the temporal progression of signal propagation offers a basis for personalized visualizations of connectivity based on CCEP measurements.

In summary, this work contributes to the field of neuroscience by providing robust methods for visualizing and analyzing brain connectivity, ultimately enhancing our ability to translate these findings into practical clinical applications.

Critically ill patients in the Intensive Care Unit (ICU) are often comatose and thoroughly monitored. Neurological complications occur in up to 20% of these patients. Therefore, monitoring of the brain, which can be performed using electroencephalography (EEG), has the potential to significantly impact the outcomes of patients at the ICU. Despite the potential of EEG as a noninvasive method for monitoring the neurological status of critically ill patients, the labor-intensive and complex nature of its application and assessment has hindered widespread implementation. Therefore, the aim of this thesis was to identify the logistical and technical prerequisites for efficient and effective neuromonitoring using EEG in the ICU.

Chapter 1 provides an overview of neuromonitoring techniques in the critically ill patient. It delves into the neurophysiological background of the EEG, the techniques used for applying the EEG electrodes, and the EEG assessment.

In Chapter 2 we present a qualitative study on the optimal conditions for EEG monitoring in the ICU. Through 12 individual and 2 focus group interviews with employees from different departments within and outside of the hospital, the current workflow regarding neuromonitoring in the ICU is identified. Additionally, we evaluate the barriers and facilitators for change in this monitoring process through the Consolidated Framework for Implementation Research (CFIR). Factors such as motivation and willingness to change serve as facilitators, while a lack of interdepartmental communication and the high workload for various healthcare professionals involved can be significant barriers.

The qualitative research reveals that the largest group monitored using EEG in the ICU consists of patients suffering from postanoxic encephalopathy, which can be a complication of a cardiac arrest. Therefore, in Chapter 3, we examine the technical requirements for optimal EEG monitoring. Specifically, we focus on the necessary number of EEG electrodes for reliable automatic classification of the EEG background pattern in postanoxic encephalopathy. By training an Random Forest (RF) classifier with input from 12, 10, 8, 6, and 4 EEG electrodes, we develop a model with a micro-averaged One-vs-Rest (OvR) Area Under the Curve - Receiver Operating Characteristics (AUC-ROC) value of 0.923, 0.924, 0.924, 0.925, and 0.923 (p-value: 0.279) for the different numbers of electrodes respectively. The constant performance of the model suggests that a reduced number of electrodes may be sufficient for monitoring this patient group, potentially reducing the workload for EEG technicians. Automatic assessment of the EEG can also contribute to a decreased workload for clinical neurophysiologists.

In Chapter 4 we provide the conclusions and future perspectives of this thesis. We have demonstrated the potential for change in the EEG monitoring workflow at the ICU of the Erasmus MC, indicating that there is an opportunity to work towards more effective and efficient neuromonitoring. Future research should focus on a broader range of logistical and technical prerequisites - including effective interdepartmental collaboration and which EEG equipment to use - thereby creating opportunities to improve treatment and outcomes of critically ill patients.

During the initial phase of diagnosis, patients with anti-NDMA-receptor encephalitis (anti-NMDARE) often experience severe symptoms that significantly impact their quality of life. Anti-NDMARE is an autoimmune disorder affecting the brain, with electroencephalography (EEG) playing a vital role in diagnosis and treatment. Identifying EEG patterns associated with positive or negative prognosis is crucial for adjusting treatment intensity. Improved understanding of diagnosis, prognosis and treatment could enhance the quality of life for anti-NDMARE patients. This thesis aimed to analyse the EEG data with Machine Learning (ML) to predict which patients exhibit positive recovery after 12 months of standard treatment.

To predict the outcome after 12 months, a Random Forest (RF) classifier was constructed using available EEG features. The EEG dataset exhibited a clustered structure due to multiple values for each patient’s EEG features. Three approaches were considered to handle this clustering: ignoring clustering, reducing clustering to independent observations, and explicitly accounting for clustering. The first two options were explored in this research. Another prominent challenge encountered early in the research was the class imbalance, which was addressed by under- and oversampling the dataset.

For the simulation sets, under- or oversampling did not yield the desired effect, as the normal sets demonstrated comparable or even superior performance compared to the the under- and oversampled sets. However, under- and oversampling improved the performance scores for the real dataset. Reducing the clusters to independent observations did not achieve high performance scores compared to ignoring clustering, both in the simulation and real data cases. Furthermore, in both cases, RF models using the EEG sets outperformed those using principal component analysis (PCA) on the clustered EEG set.

Although the performance metrics scores were not yet optimal, important features for determining class labels were identified, providing a good understanding of the dataset. Mean Decrease in Impurity (MDI) and SHAP algorithm highlighted the significance of connectivity-related features in the reduced clustering to independent observation setting. The relevance of these features became evident upon calculating the mean, minimum, or maximum. In the EEG setting, MDI emphasized the importance of the features deltapower, sampleentropy and occipital-related features. These features remain important in the reduced set. SHAP, in addition to prioritizing the same features, offered insights into how specific features contribute to the prediction of a specific observation, enhancing interpretability.

The challenges for the RF classifier in the case of anti-NDMARE are class imbalance and accurate classification of the minority class. Under- and oversampling techniques successfully improved classification of minority class observations for the original EEG set. Concluding, this set is strongly encouraged to be utilized over all sets when aiming to classify EEG features. However, this set overlooks the clustering aspect, leaving room for optimization in future research to address this limitation. Additionally, it is recommended to explore the potential of a Convolutional Neural Network (CNN) for accurate classification of raw EEG signals. Its exploration was beyond the scope of this research.

Epilepsy has been reported in 10-40% of children in the paediatric intensive care unit (PICU). Amplitude-integrated electroencephalography (aEEG), often used as neuromonitoring in the PICU, has some limitations and as a result, caretakers in the PICU may find it challenging to interpret aEEG. This may lead to uncertainty during diagnosis and increases the need for the assistance of the neurophysiology department, a costly EEG and a late diagnosis which might result in irreversible harm.
The aim of this exploratory study was to identify and evaluate observer-based features as stand-alone classifiers and combined in a random forest classifier, that could aid in the accurate classification of seizures in paediatric critical care patients with 4-electrode EEG.
Several features achieved an AUC above 0.60, with the lower border of the aEEG signal yielding the highest AUC of 0.79. The best-performing features maintained their classifying ability when tested on a larger, independent dataset. The random forest classifier did not provide better results.
Our results show that various observer-based features can aid in the identification of seizures in 4-electrode EEG. These additional features could increase the accuracy and decrease the delay and uncertainty in diagnosing epilepsy in aEEG by PICU staff, reducing the need for full setup EEG monitoring and improving patient outcomes in paediatric critical care.