Abstracts

Rationale: Electroencephalography (EEG) is frequently contaminated with muscle artifacts, electromyography (EMG). The EMG is a broad-spectrum signal. High frequency oscillations (HFOs) guided seizure onset zone (SOZ) resections have shown good therapeutic outcome. However, since HFOs stem from small, focal subsets of neurons, they are difficult to detect in conventional EEG. The EMG signals are an order of magnitude greater than HFOs and overlap frequency bands. Tripolar EEG (tEEG) attenuates EMG and consistently detects narrow-band high frequency activity (HFA) (1). Our participants performed a sequence of activities to generate muscle contamination in the tEEG to determine if muscle artifacts caused narrow-band HFA. Methods: To determine if muscle contamination in the tEEG causes narrow-band HFA, we had healthy participants (n=5) perform activities that generated muscle contamination. Sixteen electrodes were placed in standard 10-20 system locations (F7, F3, Fz, F4, F8, T7, C3, C4, T8, P7, P3, Pz, P4, P8, O1, O2) on the scalp with EEG and tEEG recorded simultaneously.  The activities performed by the subjects consisted of a series of 30 sec blocks of baseline (resting), where the subject sat still with their eyes open, and different 30 sec movements.  Each movement was preceded and followed by a baseline.  The movements (in order) included: turn head right, turn head left, chewing and swallowing, jaw clenching, blinking (10 sec fast, 10 sec slow, 10 sec using forehead), and sitting still with eyes closed. The data was acquired at 1,600 S/s (1-500 Hz, 60 Hz notch on).  We also recorded EEG and tEEG concurrently from patients (n=5) who had seizures while recording. We used time-frequency techniques to generate spectrograms. Results: Figure 1 is a typical illustration of signals recorded from a patient that had a seizure. Panel A is the spectrogram of the 14.5 min tEEG time series, Panel B. The black ellipse shows narrow-band HFA starting approximately 10 minutes prior to the seizure. Panel D is 11 sec highlighted by the small black ellipse in Panel A and Panel C is the spectrogram for Panel D. Panel D shows a nearly rhythmic pulsing. Panels E and F are 2 sec expansions of from Panel D showing HFA. Figure 2 is a typical illustration of signals recorded to analyze EMG contamination. Panel A is the spectrogram for the approximately 7 min tEEG time series of Panel B. Signals from the black rectangle spanning 11 sec are shown in Panel C which in Panels D and E are 2 sec. Panel D shows the transition from baseline to right head turn. Figure 2 Panel A does not have any narrow-band spectral patterns like those evident in Figure 1 Panels A and C. Instead the spectrograms in Figure 2 Panel A are broad band from DC to 500 Hz. Further, the time series of Figure 2 Panels D and E do not show any distinct HFA as seen in Figure 1, Panels F and G. Conclusions: HFA was detected in the tEEG of each subject that had seizures, while absent in the EEG. The EMG contamination does not generate narrow-band HFA like those found prior to seizures. Besio W., Martinez-Juarez I., Oleksandr M., Giatanis J., Blum A., Fisher R., Medvedev A., “High-Frequency Oscillations Recorded on the Scalp of Patients with Epilepsy Using Tripolar Concentric Ring Electrodes“, IEEE Journal of Translational Engineering in Health and Medicine, 2013. Funding: National Science Foundation grants 1539068 and 1430833