Apply Scout After or Before in Connectivity?

Discussions
,
1

Hi
I already computed source connectivity using Amplitude Envelope Correlation for a pair of scouts: G_front_inf-Orbital R and G_front_inf-Orbital L,
I computed it for Alpha band,

First time I selected to apply the scout After and second time I selected Before (Just to compare it)

scouts
scouts702×128 11.4 KB

This is what I got:

AEC:

Apply Scout Before: -0.054

Apply Scout After: 0.026

Why these values are so different??

What is the right value or the right way to compute AEC between scouts? using apply scout After or before?

Thanks in advance!

2

These are completely different measures, as explained in this tutorial:
https://neuroimage.usc.edu/brainstorm/Tutorials/TimeFrequency#Scouts

If the effects you are trying to observe are not robust, or your scout too large, the result might be quite random.

3

Thanks a lot!

4

May I ask, have you run into any size limitations with using the 'After' option? No matter how few scouts I use (doing this for a single participant), it takes a long time and results in an error/warning about the required space needed for this single analysis.

5

Working with full source maps can easily create gigantic matrices, especially if your epochs are long...
Does it work with the option "Before"?

6

Hi Francois,

it indeed does work fine with the "Before" option. I'm guessing I should be using that and not "After

7

What it means is that your problem comes from one of the processing steps requiring the allocation a matrix that is too large for the memory on your computer (because you have too many time points? or too many sources? or a 32bit operating system?)

I'm sorry I can't give you a clear answer about the option "before" vs. "after". We tend to think that "after" would give a better representation of the actual connectivity between all the pairs of sources within the two scouts, but it might be noisier as well. This was not evaluated formally and I can't give any reference to a publication discussing this topic.
Maybe @Sylvain, @pantazis or @hossein27en would have some references to cite?

8

Agreed with François. "After" is more computationally expensive but is expected to preserve the integrity of source time series, esp. in higher frequency bands, in the extraction of scout measures.
"Before" can be used as a proxy, if computing resources are limited, but sensitivity will be less.
A tradeoff is to first use smaller regions of interest (scouts).