[Eeglablist] Does EEG cleaning changes functional connectivity value?

[Makoto Miyakoshi]

Hi Jinwon,

The short answer is yes.
But your naive question is actually tricky and could be even misleading.
See my points below.

1. If we do not know the ground truth, how can we determine the 'signal'
(or 'noise') is changed due to some processes? The ground truth is by
definition unavailable here.
2. Only when the unwanted artifacts have perfectly linear properties that
are known to us, can we isolate them without affecting the ground truth.
This is an impossibly rare case in practice however.
3. Changed/unchanged is an extreme choice. If you ask your question in such
a qualitative way, the answer needs to be almost always 'changed'. For
example, if you import your data to EEGLAB, save it to the disk, then your
data are already 'changed' because the bit resolution of your original data
is probably different from that of EEGLAB (forced to be in a single
precision when saving). If your usual sense, you would not call it
'changed'. But technically they are 'changed'. So you have to change your
question from a qualitative one to a quantitative one to allow trivial
changes that YOU do not care.

For another example, imagine the following simple simulation: You have your
EEG data recording. I add a pure 60-Hz sine wave with a fixed level of
amplitude to your data. Now your data are contaminated with this simulated,
simplified line noise (which does not show amplitude change over
time, phase discontinuity, jitter of central frequency, etc.). You can
remove this added line noise rather well using various kinds of line noise
removal tools available. You can even use a simple notch filter to 'remove'
this artifact rather effectively. Even under these favorable situations, I
must say "functional connectivity value could be changed" according to your
criterion.

When we do signal processing, we HOPE that we reduce more noise than
signals under the situation in which ground truth is unavailable. So we
rely on simulation studies at least to make properties of our tools known
to us. Most of the tools available out there are supposed to be ok in most
of the situations. If you really care for a specific combination of data,
tools, and situations, you need to design a simulation to test the validity.

Makoto

On Sun, Apr 7, 2024 at 5:49 PM 장진원 via eeglablist <eeglablist at sccn.ucsd.edu>
wrote:

> Dear all,
>
> I'm not majoring in signaling and computer science, so I'm not familiar
> with the basic concept of functional connectivity. Can you help me?
> When we remove bad data using hand rejections or ASR clean data, we remove
> certain periods of EEG, but then, functional connectivity value could be
> changed. right? For example, because delta~gamma wave have different
> frequency range, could coupling like theta-gamma coupling and phase locking
> value be affected?
>
> Thank you,
> Jinwon Chang
> _______________________________________________
> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> To unsubscribe, send an empty email to
> eeglablist-unsubscribe at sccn.ucsd.edu
> For digest mode, send an email with the subject "set digest mime" to
> eeglablist-request at sccn.ucsd.edu
>