Question about z scores

Discussions
1

Hi,

I used the z score function in the pipeline editor on my raw source localization data and noticed that after averaging across participants that the baseline level was around .4 or .5 and someone mentioned to me that the baseline for z scores should be hovering above and below zero. I kept the ‘dynamic’ box checked…are these z scores all rectified in the positive direction?

thanks,
Andy

2

Hi Andy,

We changed recently our recommendations on the normalization of cortical sources. For a correct Z-normalization (transforming your data to a normal distribution with 0 mean and unit variance), you should not rectify your source maps before normalizing them:
https://neuroimage.usc.edu/brainstorm/Tutorials/SourceEstimation#Standardization_of_source_maps

During the baseline, you should observe oscillations around zero and with values between -1 and 1 (and not only negative!).

I would also recommend you update Brainstorm. The process "Z-score normalization" was renamed "Baseline normalization" and the option "dynamic" was removed because it caused more confusion than advantages.

Cheers,
Francois

3

Hi Francois,

I updated brainstorm and re did my z score and got the exact same normalized waveform as I previously did using the old ‘dynamic’ function. Once I had performed the z scores, I then rectified the data, and that maybe is why I’m getting an above zero (positive) baseline on my recordings, which is the rectified noise level?

thanks,
Andy

4

I updated brainstorm and re did my z score and got the exact same normalized waveform as I previously did using the old 'dynamic' function

The two versions should give exactly the same results if used correctly, they just save the files in a different way on the hard drive.

Once I had performed the z scores, I then rectified the data, and that maybe is why I'm getting an above zero (positive) baseline on my recordings, which is the rectified noise level?

Maybe. Avoid rectifying the sources explicitly, do it only for display purposes (using the option "Absolute" of the colormap).

Francois