High-Frequency Oscillations in the High-Density EEG

Epilepsy is one of the most common serious chronic neurological conditions. It is accordingly worse that ~30% of patients do not respond to medication and continue to experience seizures. For these patients, surgical intervention is the most important treatment option with a realistic hope of seizure freedom. But a good outcome, which is achieved in 60-80% of patients, is heavily depending on the precision in defining the to-be-resected brain-region This region is known as the epileptogenic zone, because it is indispensable for generating seizures. Unfortunately, there is no method to directly measure and define this area. Therefore, assessment is based on mutually complementing diagnostic approaches. In the last decade, researchers all over the world focused their interest on the question whether high-frequency electric activity of the brain could be a more accurate indicator. But this activity is typically measured invasively, which bears considerable risks. In some studies, measurement of the high-frequency-activity on the scalp was done by use of the classical electroencephalogram, i.e. 21 electrodes placed all over the scalp. Given the small-scale genesis and local propagation of high-frequencies it is likely that use of this technique misses most of the occurrences of high-frequency activity. Therefore, we will use high-density electroencephalographic recordings with 256 electrodes to detect high-frequency activity on the scalp. In addition, we want to refine and adapt currently available techniques for the automated detection of the activity of interest in the recorded data because visual detection is highly subjective and far too time-consuming for clinical practice. Moreover, the available algorithms for automated detection were developed for invasively recorded data. They need to be adapted to the characteristics of data from the scalp EEG, which suffers from a comparably low signal-to-noise ratio. Finally, pathological high frequency oscillations can easily be confounded by physiological high frequency oscillations. The only workaround so far has been to not consider high frequency oscillations from brain regions which are known to generate physiological activity in this frequency range, such as the regions being involved in memory consolidation, visual perception, or movement control. In order to better discriminate physiological from pathological HFOs, we plan to stimulate these areas with activation tasks (vision, movement, and memory). We want to find out whether the physiological subtype of high frequency oscillations reacts to this stimulation whereas the pathological does not. We hypothesize that the evaluation of high-frequency activity recorded with high-density technology on the scalp by automated detection can predict a favourable outcome better than low- density recordings and visual detection. On the long range, high-density scalp recordings could help to decide on where to position invasive electrodes before surgery. Finally, our vision is that one day scalp recordings could at least partly replace the risk-bearing invasive recordings.

Epilepsy is one of the most common serious chronic neurological conditions. It is accordingly worse that ~30% of patients do not respond to medication and continue to experience seizures. For these patients, surgical intervention is the most important treatment option with a realistic hope of seizure freedom. But a good outcome, which is achieved in 60-80% of patients, is heavily depending on the precision in defining the to-be-resected brain-region This region is known as the epileptogenic zone, because it is indispensable for generating seizures. Unfortunately, there is no method to directly measure and define this area. Therefore, assessment is based on mutually complementing diagnostic approaches. In the last decade, researchers all over the world focused their interest on the question whether high-frequency electric activity of the brain could be a more accurate indicator. But this activity is typically measured invasively, which bears considerable risks. In some studies, measurement of the high-frequency-activity on the scalp was done by use of the classical electroencephalogram, i.e. 21 electrodes placed all over the scalp. Given the small-scale genesis and local propagation of high-frequencies it is likely that use of this technique misses most of the occurrences of high-frequency activity. Therefore, in this project we used high-density electroencephalographic recordings with 256 electrodes to detect high-frequency activity on the scalp. Indeed, we could show that on average we can detect more high-frequency activity in the high-density electroencephalogram as compared to classical approaches. In addition, we refined and adapted currently available techniques for the automated detection of the activity of interest in the recorded data because visual detection is highly subjective and far too time-consuming for clinical practice. Moreover, the available algorithms for automated detection were developed for invasively recorded data and we could show that they are not appropriate for data recorded from the scalp. Finally, pathological high frequency oscillations can easily be confounded by physiological high frequency oscillations. The only workaround so far has been to not consider high frequency oscillations from brain regions which are known to generate physiological activity in this frequency range, such as the regions being involved in memory consolidation, visual perception, or movement control. In order to better discriminate physiological from pathological high frequency oscillations, we stimulated these areas with activation tasks (vision, movement, and memory). However, we found that this strategy does not help in distinguishing physiological from pathological HF high frequency oscillations Os. In conclusion our negative results fit well into the recent claim of many scientists that the initial hype about high frequency oscillations is not supported by data if strict scientific criteria are applied in research.