Weird Power Spectral Density

I am looking at the power spectral density of the preproc_bold.nii.gz files and am getting some weird results.

Usually the PSD of a BOLD fMRI signal looks something like this: (log-log)

Where we have a high power at low frequency, and a low power at high frequency

But the PSD for the dHCP preproc_bold.nii.gz files look like this:
dHCP_PSD

You can also see a similar thing on FSLeyes in a single voxel using view → power spectra, and then adjust to log-log plot

This is the opposite of what I usually see in the BOLD fMRI signal… like mirror image along a vertical axis… (flipped left to right w.r.t. frequency)

Is this due to something that was done in the preprocessing?
Anyone have any insight?

Cheers

Maybe I am completely missing the point, but since your frequencies are also negative, then it makes sense that it’s symmetric, and once you flip, it looks ok.

Oh sorry Daniele, I should have made it clear the bottom is log frequencies
So negatives on the x-axis correspond to frequencies < 1, but > 0

So -1 on the x-axis is a frequency of 0.1 (i.e. 10^(-1) = 1/10 = 0.1)

MANAGED BY INCF