Computational protocol: Eye blink correction: a test on the preservation of common ERP components using a regression based technique

Similar protocols

Protocol publication

[…] The emotion induction go/no-go task that was used in the present study was presented using E-Prime software (Psychological Software Tools, Pittsburgh, PA). In standard Go/nogo paradigms, participants are required to press a button as fast as possible given a particular category of stimuli (the Go condition, 67%) and withhold responding given another category of stimuli (the Nogo condition, 33%). Participants in this study were instructed to click the button for each letter presented but to avoid clicking when a letter was repeated a second time in succession. A detailed description of the task can be found in . [...] EEG was recorded using a 128-channel Geodesic Sensor Net and sampled at 250 Hz, using EGI software (Electrical Geodesics, Inc., Eugene, OR). Data acquisition was started after impedances for all EEG channels were reduced to below 50 kOhm, which is acceptable with high input impedance amplifiers (; ). All channels were referenced to Cz (channel 129) during recording and later rereferenced against an average reference (; ). Data were filtered off-line using a 1–30 Hz finite impulse response (FIR) bandpass. Correct no-go data were segmented into epochs from 400 ms before to 1000 ms after stimulus onset.From this point forward, the data was split into three groups for the purposes of our analysis. To create realistic research circumstances, it’s important to note that the artifacting parameters mentioned were set a priori, without any known intention, at that time, whether this comparison was ran:The uncorrected no blink data- This was ‘clean’ data, without any eyeblinks, for which no correction method was used. This data followed our standard artifacting procedure: channels were automatically marked bad when they exceeded a transition threshold of 150 µv over the entire segment (max-min). Remaining eye blinks were detected when the vertical eye channels exceeded a threshold of 150 µv (max-min) within a 160 ms (moving) time window within each trial after running a 20 ms moving-average smoothing algorithm across the entire trial period. Eye movements were detected when horizontal eye channels exceeded a threshold of 100 µv (max-min) over a 200 ms time window. Furthermore, each segment of the EEG was excluded from averaging if 20 or more channels were rejected. These settings were determined by extensive tests on a sample of the data that yielded the best artifact detection for our data. In addition, all segments were visually inspected by a trained research assistant blind to the hypotheses. After the artifacting, bad channels were removed using spherical spline interpolation.The corrected no blink data – This was the same ‘clean’ data used for the uncorrected no blink data (described above), with the exception that the correction method was applied. First, the data followed the exact same artifacting procedure as described above, after which bad channel replacement was run. Then, the eye blink correction tool was run (setting the blink slope threshold at the recommended 12 µV/ms), followed by another, somewhat stricter, round of artifacting and subsequent bad channel replacement. This round of artifacting was mostly meant to remove anomalies in the data that were leftover by the correction data. This amounted to the rejection of less than 5% of the total trials.The corrected blink data- This data contained eye blinks that were corrected using the above-mentioned procedure using the Gratton correction algorithm. These were data from the same participants used in the no blink data.After these procedures, data were averaged, average referenced, coded, and baseline corrected for the 400 ms preceding the stimulus onset. The coding was performed by research assistants blind to the hypothesis of the study. Five common stimulus-locked ERP components, such as the N1, P2, N2, P3p (parietal P3), and the frontal P3, were derived when subjects correctly inhibited on a nogo trial using the traditional peak picking method. The N1, P2, and N2 were coded as maximum positivities or negativities before 500 ms after the stimulus had appeared, whereas the P3p and P3 were typically coded 500 ms post stimulus (to max of 900 ms). Data were analyzed according to standard lab procedures (for more details, see ) using IBM SPSS statistics software (version 20). Outliers were removed if they were two standard deviations above the mean. In the current analyses, no more than 3 outliers were removed in total. […]

Pipeline specifications

Software tools E-Prime, SPSS
Applications Miscellaneous, Neuropsychology analysis
Diseases Eye Abnormalities