jonescompneurolab / spectralevents Goto Github PK

Based off of the discussion in #9.

This should reproduce the approach presented in Shin et al. 2017 Fig. 4 by passing in a stack of TFR trials and plotting the correlation coefficient between % suprathreshold power samples and avg power at different values of FOM threshold.

The timing of the peak spectral event latency for spectral events quantified from demo data with the MATLAB versus Python version of the toolbox are offset by 1.66667ms (MATLAB events being exactly 1 sample point later at 600Hz). (Suspect but am not 100% certain that this discrepancy may partially have something to do with differences in decimal point precision between MATLAB and Python, as functions like np.arange on line 198 of spectralevents.py return 4 fewer decimal points which could contribute to a difference in timing - ex: 0.1767 vs. 0.17666667.)

README.md currently only discusses the matlab functions.