Phase locking value

Discussions
,
1

Hi all,

In the Brainstorm I am calculating connectivity measures using Phase locking Value (PLV) for N*N at selected time window (-200, 1000).

Some notes recommend to use baseline normalization after computing PLV metric (for instance, https://praneethnamburi.wordpress.com/2011/08/10/plv/)

Unfortunately, the attempt to do it failed each time, and the error message appears in Brainstorm “Cannot process values averaged by time bands”.

I have two questions: is it really necessary to compute baseline normalization for PLV and, if so, how it could be possibly done in Brainstorm?

Thank you,

Oksana.

2

Hello

Could you describe better what you are doing?
How do you compute your PLV files? What process do you call to get this error?
The PLV process should not create files with time bands in them… If you are explicitly compute time bands, just skip this step.

Thanks
Francois

3

Hello Francois,

First I computed PLV for N*N (using a magnitude measure) for the epoch (-200, 1000 ms).

1.jpg755×145 22.5 KB

Then I chose in the "standartize" menu the option "baseline normalization" using the baseline (-200, -1 ms).

2.jpg561×784 84.4 KB
3.jpg599×715 110 KB

After that the error message appeared "Cannot process values averaged by time bands".

Please also find attached the screenshots of the process.

I am also wondering, if the normalization is necessary in case of PLV measure, because not all recommendations are refer to do it?

Thank you,
Oksana.

4

This process “Baseline normalization” is designed to work for a file that has a time definition, as explained here:
http://neuroimage.usc.edu/brainstorm/Tutorials/TimeFrequency#Normalized_time-frequency_maps

In your PLV file, there is no time dimension anymore, this file is indeed considered by Brainstorm as one value in one time band. If you want to normalize with respect to a baseline after computing the PLV value, it would mean something like dividing the PLV value post-stim by a PLV value pre-stim or from a baseline condition. Or running a non-parametric test between an active condition and a control condition. You need to identify clearly the measure you want to use.

5

Thank you very much for the suggestion!

6

Hi

I'm interested in examining whole brain connectivity using PLV. Specifically, looking to examine PLVs with a 1Hz frequency resolution. Is there a built in method to calculate complex values using a morlet transformation rather than Hilbert?

Thanks

7

Specifically, looking to examine PLVs with a 1Hz frequency resolution. Is there a built in method to calculate complex values using a morlet transformation rather than Hilbert?

No, this is not an available option .
The new Envelope Correlation process is the only one that includes an option to chose between Hilbert and wavelets.

This is possible, but you would need to code it yourself from a Matlab script, after reading the signals from the database.
https://neuroimage.usc.edu/brainstorm/Tutorials/Scripting