Concurrent EOG and high-impedance EEG was recorded with a Biosemi ActiveTwo with Actiview software (Minimitter, Bend, OR) 64-channel EEG system, in accordance with the international 10–20 system (Neymotin et al., 2020), located in an electromagnetically shielded room while participants performed the MGS task (described above). The task stimuli were presented on a computer 80 cm from the subject, and head position was maintained using a chin rest. Initial data were sampled at 512 Hz and down sampled to 150 Hz during preprocessing. Data were referenced to external electrodes corresponding to the mastoids due to their proximity to the scalp and low signal recording. An initial bandpass filter was set to 0.5–75 Hz. We used a revised version of an established protocol (Makoto Pipeline) for preprocessing compatible with EEGLAB (Delorme and Makeig, 2004). This protocol removes flatline channels (maximum tolerated flatline duration: 8 s), low-frequency drifts, noisy channels (defined as more than 5 standard deviations from the average channel signal), large amplitude artifacts, and incomplete segments of data (Jung et al., 2000; Hipp and Siegel, 2013). Deleted channels were replaced with interpolated data from surrounding electrodes. Continuous signals were divided into epochs, time-locked to stimulus onset (∼1–3 s). Data epochs were cleaned with an amplitude threshold of −500 to 500µV to exclude eye blinks. The resulting data were referenced to the average of all electrodes. As a final preprocessing step, independent component analysis (ICA) was performed to identify eye-blink artifacts and remove their contribution to the data. We confirm that gamma bursts are not present due to eye movement by plotting the spectrograms from the EOG channels.

Please refer to the LNCD Github Repository
Note: you need the hera function for the pipeline to run

• run_preprocessing_pipeline.m
  • remark.m function pulls the raw data from the database and remarks the trigger vales to single digits corresponding to the appropriate epoch. Also see EEG Remarking Status Channel.
• preprocessing_pipeline.m
  • Loads in the desired subject
  • Loads in the file containing the coordinates of the cap electrodes
  • Checks to see if the subject has already been preprocessed. If yes, skips and moves on to the next
    • file_locs.m will load in all the path and file names
  • References the data to the electrodes corresponding to the mastoids
  • Filters out the data below 0.5 Hz (low frequency drifts) and 70 Hz
  • Resample the data to 150 Hz (all tasks besides SNR) or 512 Hz (for SNR task)
  • Removes external channels
  • Imports channel coordinates into the EEG file
  • Bad channel rejection
    • See clean_rawdata.m EEGLAB plugin
    • The following criteria was used:
    • arg_flatline: 8
      • Maximum tolerated flatline duration. In seconds. If a channel has a longer flatline than this, it will be considered abnormal.
    • arg_highpass: [0.25 0.75]
      • Transition band for the initial high-pass filter in Hz. This is formatted as [transition-start, transition-end]
    • arg_channel: 0.7
      • Minimum channel correlation. If a channel is correlated at less than this value to a reconstruction of it based on other channels, it is considered abnormal in the given time window. This method requires that channel locations are available and roughly correct; otherwise a fallback criterion will be used.
    • arg_noisy: 5
      • If a channel has more line noise relative to its signal than this value, in standard deviations based on the total channel population, it is considered abnormal.
    • arg_burst: 15
      • Standard deviation cutoff for removal of bursts (via ASR). Data portions whose variance is larger than this threshold relative to the calibration data are considered missing data and will be removed.
    • arg_window: 0.3
      • Criterion for removing time windows that were not repaired completely. This may happen if the artifact in a window was composed of too many simultaneous uncorrelated sources (for example, extreme movements such as jumps). This is the maximum fraction of contaminated channels that are tolerated in the final output data for each considered window.
  • Interpolate missing channels
    • See pop_interp.m
      • Dataset includes a few subjects that used a 128 cap as opposed to a 64 channel cap. The code removes the 4 channels that are found in 128 but not 64 and reinterpolates the missing
  • Reference data to average for ICA
• runICAs for eye movement
• Remove bad epochs via epochclean.m
  • Uses pop_epoch.m
• Move channel locations if subject used a 128 channel cap so that the coordinates are the same as the 64 channel cap
  • See homogenizeChanLoc.m
• Filter out the 60 Hz artifact
• Fully preprocessed data will be in Hera/Projects/7TBrainMech/scripts/eeg/Shane/preprocessed_data/SNR/AfterWhole/ICAwholeClean_homogenize

• Initial pipeline help from Alethia in /Volumes/Hera/Projects/7TBrainMech/scripts/eeg/Alethia
• Taken over by Shane /Volumes/Hera/Projects/7TBrainMech/scripts/eeg/Shane