EEG recordings during resting-state and the maintenance periods of a spatial working memory task in humans

Scripts used to analyze data for the manuscript submitted for publication in EJN

Script_Curve_Fitting_HBM.rtf

Dr. Hadj Boumediene Meziane: [e-mail] 

We therefore considered this continuous change in power as an extraneous variable yk(x) impacting the measured power spectrum Pow(Ek), and modeled it with a binomial equation that best fit the data, where the coefficients in pi are in descending powers, and the length of p is (n+1), k is trial number (k = 1 to 10):

yk(x) = p1 . x2 + p2 . x + p3

In order to statistically compare the topographies between the trials with perfect recall and the trials with failed recall, we subtracted this variable from the mean spectral topographies of each subject and for each electrode by first producing the mean spectral curves of each maintenance trial in the theta and alpha frequency bands, taking into account the IAF, and then calculating the coefficients (p1, p2 and p3) of the binomial equation using the Matlab function polyfit.m. Once the coefficients were determined, this estimate was subtracted from each power spectrum matrix using the following formula:

PowFit(Ek) = Pow (Ek) – yk(x)

Script_Perf_Fail_EEG_Power_Spec_HBM.rtf

Dr. Hadj Meziane: [e-mail] script calculates EEG power spectra then compares perf and fail conditions, then plots brain topographies with statical results

Script_Perf_Fail_EEG_Sources_Spec_HBM.rtf

Dr. Hadj Boumediene Meziane: [e-mail] script compares EEG source spectra then compares Perf vs. Fail conditions then plot statistical results (significant voxels) on MRI volume