Electrodes Size Impact on Intracranial EEG

The in vivo portion of this study was approved by the Institutional Review Board of the Children’s Hospital of Orange County (CHOC). Informed consent was obtained prior to involvement in the study. Three human subjects with medically intractable epilepsy were each implanted with a high-density 8 × 8 subdural grid of intracerebral EEG electrodes (Ad-Tech FG64C-MP03X-000) in the clinically determined seizure onset zone (SOZ) as part of phase 2 pre-surgical invasive monitoring. Patient information is given in table 1. Each electrode had an exposed surface area of 1.08 mm² and electrode spacing was 3 mm center-to-center (this is the same grid used for the in vitro experiment in section 3.1, so we will similarly refer to these as the ‘small’ electrodes). The effective surface area was changed by electrically shorting adjacent electrodes in groups of two and four, thereby mimicking larger surface areas of 2.16 mm² (‘pair’ electrodes) and 4.32 mm² (‘quad’ electrodes), respectively (figure 3(A)). This was done by connecting jumper wires to the paired electrodes at the jack box outside the patient’s body (figure 3(B)). The jack box combines the individual electrode wires into an integrated cable before connecting to the amplifier. A quick-release connector enabled rapid reconfiguration of the electrode shorting, minimizing disruption to the patient’s recording. The jack box and jumper wires were placed in a Faraday cage to minimize electrical interference.
We collected 20 min iEEG recordings for each of three different electrode surface areas (small, pair, quad) from a grid in a static brain location while the subjects were sleeping. The sampling rate was 5 kHz, and the data were referenced to the common average of the 8 × 8 grid. The iEEG data were high pass filtered using a zero phase FIR filter at 1 Hz and notch filtered at 60 Hz, 120 Hz, and 180 Hz to remove electrical line noise before analysis. All analysis was done using custom code in MATLAB 2018b.
Similar to the in vitro study, we compared the results to the theoretical circuit model by generating ‘simulated pair’ electrode signals (consisting of the mathematical average of adjacent pairs of small electrodes) and ‘simulated quad’ electrodes (the mathematical average of four adjacent small electrodes). The averaging of signals was done on the raw iEEG data and the data were then re-referenced and filtered as described above. We compared the pair and quad electrode recordings to their associated simulated signals.