Prospective validation of electroencephalography-correlated functional MRI (EEG-fMRI) in the presurgical work-up for epilepsy surgery.

Epilepsy is a frequent neurological disorder with a prevalence of around 0.5%. One third of patients have epileptic seizures that are not controlled with antiepileptic drugs. These refractory seizures have a negative physical, psychological and social impact. Patients with seizures starting from one circumscribed area in the brain (focal epilepsy) can be candidate for epilepsy surgery. The goal of the presurgical evaluation is to define the epileptogenic zone, the part of cortex that is indispensable for the generation of seizures, and to assure it can be safely removed or disconnected. A unique non-invasive tool to localize this region is lacking. The combination of different modalities can more or less indicate the localization and extent of the epileptogenic zone, but each technique has its limitations.The aim of my PhD work was to investigate the added value of simultaneous electroencephalography and functional magnetic resonance imaging (EEG-fMRI) in this diagnostic process. Epileptic discharges (seizures and spikes) were registered by EEG and their associated hemodynamic changes by fMRI. By combining both modalities, our goal was to localize the source of this pathological activity.In a first step, the technique was implemented in the clinical setting. Crucial was the integration of an EEG-electrode system, suitable for long-term video-EEG recordings and compatible with MR-imaging. Good quality EEG-fMRI measurements could be achieved in a safe and patient-friendly way. Before EEG-fMRI could be used for clinical purposes, knowledge about reliable interpretation of the results was required. Sensitivity and specificity of interictal spike-related EEG-fMRI for detecting the ictal onset zone was evaluated in a clinical validation study, using 21 well-characterized patients. We determined the appropriate statistical thresholds to achieve the optimal diagnostic performance. Hemodynamic changes, related to focal epileptic spikes, appeared to have widespread localizations. There are reasons to assume that EEG-fMRI visualizes an epileptic network, rather than a circumscribed irritative zone. The part of the network, which contained the maximal significant activation, was a sensitive and specific marker for detecting the ictal onset zone.EEG-fMRI processing requires the visual detection and manual marking of epileptic spikes on EEGrecorded during fMRI. This procedure was very time-consuming and subjective and could compromise the implementation in clinical practice. We, therefore, developed a semi-automatic analysis method, based on an EEG-template of patient-specific epileptic spikes recorded outside the scanner. This template was correlated with the EEG recorded in the scanner. A template-specific correlation threshold was defined in healthy subjects. Time points in the patient, with suprathreshold correlations, were used to build the fMRI model. Our method was time-saving and straightforward compared to conventional visual methods, with preservation of diagnostic performance. Epilepsy is increasingly recognized as a network-disorder, in which focal discharges can, through propagation, influence distant brain areas. Knowledge about these networks may contribute to a better understanding of epilepsy-associated cognitive morbidity. Moreover, network-nodes could potentially offer new targets for tailored anti-epileptic interventions (like deep brain stimulation). In a study of 28 patients with focal epilepsy, hemodynamic changes, related to seizures and epileptic spikes were compared. Ictal SPECT (single photon emission computed tomography) visualizes changes in perfusion during seizures and interictal EEG-fMRI alterations in tissue oxygenation during epileptic spikes. These hemodynamic changes fluctuated in the same direction and within a common spatial network. The overlap was not limited to regions of epileptic discharge origin, but could be found in patient-specific brain areas at a distance. Some patients unexpectedly experienced a seizure during EEG-fMRI. Heterogeneous temporal fluctuations in BOLD functional connectivity with the ictal onset zone were found in the preictal phase. These changes in connectivity may reflect the recruitment of seizure-facilitating or the loss of seizure-inhibiting brain interactions. We conclude that the ictal onset zone can be localized in a sensitive and specific way, using interictal EEG-fMRI. Spike-related EEG-fMRI studies should be interpreted with a network concept in mind. A more automated processing will contribute to the use of the technique in a clinical setting. The necessary conditions are now met to start clinical utility studies. Future methodological challenges involve the analysis of spike-negative EEG-fMRI. Data-driven fMRI techniques may offer a solution in these cases.status: publishe