Project Lead: Shane McKeon
Project Start Date: September 2023
Current Project Status: Completed
Datasets: LNCD 7T
Github Repository: https://github.com/LabNeuroCogDevel/Cortical_SNR_Development
Publication DOI: https://doi.org/10.1016/j.pneurobio.2024.102695
Conference Presentations
S. McKeon, F. Calabro, M. Perica, B. Luna. Reliability of cortical signal processing is driven by glutamate maturation and supports working memory development. Flux Society, Paris, France. September 2022. (Graduate Poster/Abstract).

S. McKeon, F. Calabro, M. Perica, B. Luna. Reliability of cortical signal processing is driven by glutamate maturation and supports working memory development. Society for Psychophysiological Research, Vancouver, Canada. September 2022. (Graduate Poster/Abstract).

• Preprocessing
  • Preprocessing can be run using 01_Cortical_SNR_Preprocessing.sh

  •  matlab -nodesktop -r "addpath(genpath('../Preprocessing_Functions/')); run_preprocessing_pipeline('SNR')" 

    • Pulls in raw data from hera('Raw/EEG/7TBrainMech')
    • run_preprocessing_pipeline.m (/code)
    • Set task as 'SNR'
    • Bandpass filter between 0.5 Hz and 70 Hz
    • Downsamples the data from 1024Hz to 512 Hz
    • Removes bad channels (the following criterion is 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.
      • Interpolates missing channels
        • 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 channels that were removed from above
    • Run ICA to identify eye movements and blinks
    • Homogenize Channel locations
      • Read in the channel locations and make sure all files have the correct locations, especially the few subjects who were ran using a 128 channel cap
    • Filter out 60 Hz artifact from line noise
    • Fully preprocessed data will be in Hera/Projects/7TBrainMech/scripts/eeg/Shane/preprocessed_data/SNR/AfterWhole/ICAwholeClean_homogenize

• Calculate Total, Evoked, and Spontaneous Activity
  • Run the bash script 02_CreateIndividualSubjectFiles_EvokedSpontaneous.sh

  •  matlab -nodesktop -r "addpath(genpath('code/')); totalEvokedSpontaneous('4')" 

    • Calls totalEvokedSpontaneous.m (Cortical_SNR_Deveopment/code)
    • Input: triggerValue
      • 2: 20 Hz condition
      • 3: 30 Hz condition
      • 4: 40 Hz condition
    • This code will output a single csv file for each subject with their total, spontaneous, and evoked activity for every electrode

• Combine Individual Subject Files
  • 03_CombineSubDataframes.R

  • #!/usr/bin/env Rscript
    source("code/CombineSubjectDataFrames.R")
    combineSubjectDataframes(20)

    • Reads all of the individual subject csv files and combines into one giant csv
    • Set working directory (/Volumes/Hera/Projects/7TBrainMech/scripts/eeg/Shane/Results/SNR/TEI_indivSubs_allChans/)
    • Set file pattern you want to look for
      • 20Hz, 30Hz or 40Hz
    • This will save a csv file for all subjects for each stimuli condition

• Combine All Stimulus Conditions
  • 04_combineAllStimConditions.R
    • Combines the csv files on the previous step into one large csv files with all frequency stimulus conditions
    • Preforms outlier detection (2 SDs above the mean)

• Impute missing data and Calculate frontal region PCA
  • 05_createImputed_PCAdataframes.R
    • Read in csv with all subjects info
    • Select the freq and hertz combo you want to analyze
      • Hertz: the stimuli they heard (either 20, 30, or 40)
      • Freq: the freq value you want to see their activity at
    • Calculates SNR: Evoked/ Induced
    • Runs imputation on the missing data from outlier detection for SNR, evoked, spontaneous, and total individually
      • mice imputation package
    • Selects channels you want to include in the PCA
    • For this project we chose the frontal electrodes ('F3', 'F5', 'F7', 'F1', 'F2', 'F4', 'F6', 'F8', 'AFz', 'AF1', 'AF2', 'Fp1', 'Fp2','Fz', 'AF5', 'AF6')
    • Run PCA using prcomp for SNR, evoked, spontaneous, and total individually
      • Each one is saved out with the corresponding channel location data with name of measure (evoked, induced, etc)chanlocs.csv (/Volumes/Hera/Projects/7TBrainMech/scripts/eeg/Shane/Results/SNR/)
    • Combine PCA components 1, 2, and 3 for all measures and save out ('/Volumes/Hera/Projects/7TBrainMech/scripts/eeg/Shane/Results/SNR/SNRmeasures_PCAvalues.csv')

• Figures and Statistics
  • 06_SNR_Results.Rmd
  • To assess developmental trajectories of cortical SNR activity, we implemented GAMMs on the first principal component, PC1, of evoked power, spontaneous power, and SNR, including random intercepts estimated for each participant.
  • Regression splines were implemented (4 degrees of freedom) to assess linear and non-linear effects.
  • Auditory measures that were found to significantly change across adolescence were then used to test for associations with our MRSI measures, glutamate (Glu), GABA, and Glu GABA Asymmetry using linear mixed effect models (lmer function, lme4 package in Rstudio).
  • We first tested for significant main effects of the auditory measure on the MRSI parameter while controlling for age and hemisphere (left or right DLPFC).
  • We additionally tested for auditory measure-by-age interactions while controlling for hemisphere.
  • We then investigated whether our auditory measures had significant associations with our working memory measures (accuracy, accuracy trial variability, response latency, response latency variability) using linear mixed effect models (lmer function, lme4 package in Rstudio).