Neural network integration from multimodal imaging in epilepsy and healthy controls
von Daniel van de Velden
Datum der mündl. Prüfung:2023-01-27
Erschienen:2024-01-17
Betreuer:Prof. Dr. Niels K. Focke
Gutachter:Dr. Andreas Neef
Gutachter:Prof. Dr. Florentin Wörgötter
Dateien
Name:Daniel_vandeVelden_full_Thesis_v3.pdf
Size:27.0Mb
Format:PDF
Description:Daniel_vandeVelden_Thesis
Zusammenfassung
Englisch
Different imaging modalities can be used to probe distinct aspects of the brain, such as cerebral anatomy (MRI), tissue diffusion directions (DTI), metabolic activity (PET), neurovascular activity (fMRI), or electrophysiological activity (EEG/MEG). The imaging modalities differ not only in terms of the aspect they represent, but also in their spatial resolution and in their temporal resolution. The medical imaging of the brain with each modality separately obtains valuable information. However, the joint use of neuroimaging techniques (measured in parallel or sequential) offers the possibility to combine the information from multiple modalities to obtain a broader picture of the human brain. Epilepsy is characterized by a permanent predisposition of the brain leading to an epileptic seizure, a transient occurrence of abnormally synchronous neuronal brain activity. The two most common epilepsy syndromes are focal epilepsy, defined by a focal origin of the epileptic seizures, and idiopathic generalized epilepsy (IGE), in which epileptic seizure activity rapidly involves both cerebral hemispheres of the brain. Epilepsy is characterized as a network disease of the brain. Changes in brain networks have been detected in patients with IGE at rest, i.e., in the absence of seizures, or discharges, and under the situation by patients completing a set specific task. Brain network changes in patients with focal epilepsy are also known and identified with respect to diagnosis and treatment options. Though it has not been established whether measuring multiple imaging modalities in parallel (EEG-fMRI) provides data of sufficient quality in the modalities to detect, for example, group differences between two cohorts or constant components in a group of subjects. This dissertation aimed at the examination and integration of brain networks from different parallel and sequential measured medical imaging modalities in cohorts of patients with epilepsy and healthy controls. In the first main chapter, the impact of inside MR-scanner measurement condition on high-density electroencephalography (hd-EEG) was investigated to assess whether known statistical group differences of EEG power and phase-based functional connectivity could be replicated in patients with IGE compared to healthy controls during a parallel acquisition of fMRI. We observed that the analysis of phased-based functional connectivity (imaginary part of coherency) of EEG data is suitable for parallel measured hd-EEG-fMRI, and that group differences in a comparison of patients with IGE against controls remain identifiable. Moreover, between thalamus and the occipital cortical brain, spatial congruence of group differences in seed-based functional connectivity (IGE>controls) was found between the two modalities measured in parallel, hd-EEG and fMRI. In the second main chapter, the influence of different inverse methods for electric source imaging (ESI), as well as different stages of interictal epileptic discharges (IEDs) on the spatial identification of the epileptogenic zone in patients with focal epilepsy was investigated. Furthermore, the spatial relationship of the [18F]fluorodeoxyglucose-positron emission tomography (18FDG-PET) hypometabolism found in patients with focal epilepsy and the different stages of IEDs (IED-onset, -rise, and -peak) were investigated. The second main chapter provides observations for the spatial least distance of ESI at the time when IEDs reach their maximum amplitude and the epileptogenic zone. In addition to that, the inverse method ‘standardized low resolution brain electromagnetic tomography’ (sLORETA) was observed to provide best results in identifying the epileptogenic zone. In the third main chapter, the presence of metabolic, vascular, and neuronal resting-state networks, spatially stable across a cohort of healthy subjects and patients with focal epilepsy, were investigated in a fully simultaneous hd-EEG/fMRI/18FDG-PET dataset. The possibility of identifying known resting-state functional networks from simultaneous hd-EEG/fMRI/18FDG-PET data using group independent component analysis in each modality was demonstrated. Differences in the spatial expression of these networks were observed among modalities and could reflect differences between modalities in regard of signal origin, as well as spatial and temporal resolution. Overall, this dissertation provides results from the integration of networks from different imaging modalities, from a parallel or sequential measurement setting. Identifying relevant group differences between IGE and controls from an hd-EEG with parallel measured fMRI linked our knowledge of network changes in IGE across modalities. Furthermore, it encourages future EEG-fMRI epilepsy studies to subject both imaging modalities to same analyses and consideration of their results for conclusions. Further work in this dissertation promotes careful choice of parameters in ESI in pre-surgical assessment in patients with focal epilepsy. Furthermore, it helps to elucidate the relationship between electric source imaging of interictal epileptic discharges and 18FDG-PET in preoperative diagnosis and provides an impetus to investigate this relationship in future work. Finally, we utilized the benefit of temporal synchrony of the fully simultaneous measurement of three modalities (hd-EEG/fMRI/18FDG-PET) to demonstrate the presence and spatial concordance of functional resting-state networks in three modalities and provide evidence for the rich potential of this measurement set-up. Across projects, this dissertation demonstrates that integrating imaging data from multiple modalities can provide broader insight into the objects of study in neuroscience and, more broadly, neurological disease (exemplified here by epilepsy).
Keywords: EEG; fMRI; neural networks; FDG-PET; epilepsy