Abstracts

Rationale: Abnormally excessive and synchronous neuronal brain activity is an inherent characteristic of major number of seizures. Therefore, detection and localization of the onset of this ictal synchronous activity recorded in MEG and EEG modality are critical for the epilepsy patient's treatment outcome. There is a great need for multimodal (EEG and MEG) methodological approach addressing the synchronous changes across the frequency bands as an index of seizure onset. Our work is an effort to improve the detection and localization of seizure onset by investigating each modality in synchronized changes in the frequency power across delta, theta, alpha, beta, gamma, and high gamma bands during the time interval preceding the seizure onset. Methods: Clinical case: Three typical ictal events were recorded from 30 yo epilepsy patient during outpatient MEG exam. Waveform analysis of the ictal onset was performed by using conventional Equivalent Current Dipole (ECD, Neuromag) method. For frequency power analysis the ictal onset was selected based on the EEG and MEG latency. In addition to ECDs analysis, for calculating the source space induced power, a continuous Morlet wavelet transform was used on four 6s windows (pre-ictal 4s, ictal-2s) a baseline and three windows with ictal events. The dynamic statistical parametric mapping (dSPM) inverse operator was used for the source estimation. The power was transformed into dB, and z-scores were calculated using the baseline's mean and standard deviation. For each window, the top positive and negative z-scores were selected (above the 95 and below the 5 percentiles, respectively). This pipeline was created as part of the open-source Multi-Modal Visualization and Analysis Tool (mmvt.org) [N. Peled and O.Felsenstein et al. (2017). MMVT - Multi-Modality Visualization Tool. DOI:10.5281/zenodo.438343]. Results: Figure 1 shows increases the power of frequency between 1-20 Hz and High Gamma (70-120 Hz) at the ictal onset (0-50 ms) as compared to baseline (Z=4,  p=0.0005, Z=3.3, p=0.001, respectively) for Ictal #1. It was also observed the decrease of the delta power prior to ictal onset (-1s to 0s) (Z=14, p<0.00001) for the Ictal #1 data. Similar results were observed in Ictal #2-3 in MEG modality but not in EEG (see Figure 1).  Figure 2 shows the results of the waveform corresponding to Ictal #1. The ECD localization results suggested the left posterior insula and left lateral sulcus for ictal #1 onset. Localization of the delta power increase is showing the similar region of activation corresponding to the Ictal #1 onset (see Figure 2 B and C).   Conclusions: Our new developed pipeline for frequency analysis of spontaneous EEG and MEG data shows promising results of the spatial-temporal dynamic of synchronous changes at the ictal onset across three seizures recorded during the MEG exam. The localization of the frequency power available in our pipeline might improve the accuracy of seizure onset localization. Our preliminary results are suggesting the superiority of the MEG modality in the spatial-temporal dynamic of seizure onset as compared to EEG modality.  Funding: NIH grant S10RR031599