High Frequency Oscillations in Epilepsy: Towards Clinical Application

Abstract

In patients with a focal source of epileptic seizures, epilepsy surgery can be a treatment option. The location of the epileptogenic brain region is determined by various non-invasive imaging and neurophysiological recordings, including electroencephalgorphy (EEG) and magnetoencephalography (MEG). In some patients invasive EEG is used to pinpoint the focus even more precisely. These recordings are traditionally searched for so-called spikes, whose location is an indicator for the epileptogenic zone. High frequency oscillations (HFOs), divided into ripples (80 – 250 Hz) and fast ripples (250 – 500 Hz), might be a better marker for the epileptogenic zone than spikes. Retrospective research shows that their location is related to the seizure onset zone, and that removal of brain areas that generate HFOs is related to increased chance of seizure freedom. The aim of this thesis was to move forward to clinical application of HFOs, by improving detection and identification of HFOs in both invasive and non-invasive recordings, by determining the value of HFO analysis for identification of the epileptogenic zone and by exploring the clinical opportunities of HFOs in non-invasive recordings in non-surgical patients with (suspected) epilepsy. We automatically detected ripples and fast ripples in intra-operative electrocorticography (ECoG) before and after surgical resection of the epileptogenic focus and showed that the amount of spikes, ripples and fast ripples all decrease after the resection, irrespective of the outcome. The presence of fast ripples after the resection predicted seizure recurrence, especially when they were present already before the resection. We showed that ripples preceded spikes in non-invasive EEG, indicating that they are different phenomena, promoted by certain brain states. In children with Rolandic spikes in EEG, who are not eligible for surgery, the presence of several ripples was related to the number of seizures a child experienced. We identified ripples in MEG similar to EEG by the use of beamformer virtual sensors. By using an automatic detection algorithm ripples were detected in 64% of the patients, usually in the location of the expected epileptogenic zone. We used the same beamformer virtual sensors to increase the yield of ripples in EEG. Ripples related to the epileptogenic zone in EEG were also found in patients without spikes, and simultaneously recorded MEG and EEG showed they are complimentary techniques, with only 24% of the ripples occurring simultaneously in both techniques. Our findings show that the use of HFO analysis has benefits in epilepsy surgery, although caution is advised with respect to physiological HFOs. Repeated recording of intra-operative ECoG can identify patients who are at risk of seizure recurrence. Analysis of ripples in EEG and MEG during presurgical workup can confirm a suspected epileptogenic region, and might be used to evaluate seizure activity. Further proof of the added value of HFO analysis above spike analysis will encourage more people the perfo