303.Poster session. Oscillations: EEG.
Monday, Nov 17, 2014, 8 AM-12 PM.
I really enjoy poster sessions and the unique opportunities for interacting with the authors that they offer. This morning, I spent some time at the session on brain oscillations as studied by EEG. Here are highlights from a couple of posters (three, actually) that I had the opportunity to discuss with their authors. My apologies to the presenters whose posters I did not cover. Note that any inaccuracy or outright misunderstanding in what follows is my responsibility alone!
303.03/C39. Phase dependency of long-range neuronal transmission in entrained neuronal networks: A combined tACS-TMS-EEG study. K. D. FEHÉR, Y. MORISHIMA.
303.05/C41. A method for removing tACS artifacts from EEG data. Y. MORISHIMA, K. D. FEHÉR.
The authors of these posters, from the University of Bern, Switzerland, have combined seemingly for the first time transcranial alternating current stimulation (tACS), a noninvasive approach to modulate brain rhythms, together with transcranial magnetic stimulation (TMS), which allows sending brief, sudden pulses of simulation to the cortex, and EEG recordings.
First of all, they had to find a way of removing the tACS artifact that was orders of magnitude larger than the actual brain signals in the EEG (poster C41). Their approach involved upsampling of the EEG signals so that the EEG sampling rate was a multiple of the tACS frequency, trigger timing adjustment , then moving-window average filter, and finally PCA. Using that method, the authors were able to retrieve clean EEG and visual evoked potentials.
The authors then investigated how the phase of ongoing brain oscillations, here imposed by the tACS (6 Hz delivered to both frontal and parietal areas using 2 stimulators), influenced brain responses to sudden, punctual stimulation delivered by TMS (poster C39). They found widespread dependence of TMS-evoked potentials on the phase of tACS when both frontal and partial cortex were stimulated in phase. Interestingly, effects were restricted to the posterior electrodes when the phases of stimulation of frontal and parietal cortex were opposed.
Their preliminary results confirm the importance of ongoing brain oscillations in modulating brain responses. The next steps according to the authors will include testing other frequencies of tACS, such as the gamma band.
303.23/C59. An automated seizure onset zone detector using high frequency oscillations. S. GLISKE, W. C. STACEY.
The authors of this poster, from the University of Michigan, were interested in high-frequency oscillations (HFOs), brief periods of oscillatory activity (generally between 80 and 200 Hz) recorded by intracranial EEG in patients suffering from severe epilepsy. It is thought that HFOs are a specific marker of the seizure onset zone (SOZ), the part of the brain where seizures originate from and that should be removed surgically to cure the patients from their epilepsy.
HFOs are hard to detect “visually” by browsing the EEG; therefore, algorithms to detect HFOs have been developed, but perform poorly because there are many “false” detections. The authors therefore developed a series of algorithms that would detect and label those artifacts, avoiding false positives. They tested their algorithm in clinically annotated intracranial EEG recordings from the intracranial EEG portal, a data-sharing initiative (ieeg.org), as well as in data from patients at their local institution.
They found that their automated algorithm detected localized HFOs in just over half of the patients (53%). In all those cases, the HFOs were included either in the SOZ, as labeled by expert clinicians, or in the surgically resected area. These two zones are collectively considered the gold standard for localizing the source of seizures in epilepsy surgery. Importantly, although the current study was retrospective, the algorithm is fast enough that it can be run in real-time as data are collected.
In conclusion, the authors presented a specific, albeit not very sensitive, approach to localizing the seizure onset zone using an automated high-frequency oscillation detector. The next steps according to the authors will include investigating whether HFO detection influences clinical decision making. The possibility to perform data analysis online will prove crucial in that respect.