[Eeglablist] Discrepancy in data output between different .mff to .mat conversion techniques.

[Dylan Vlasak]

Good afternoon,

I have a question about voltage differences that I observe when converting
from .mff --> .mat using EEGLAB versus NetStation Tools v. 5.4.3-R. The
newest version of NetStation Tools we run (5.5) no longer supports .mat
conversions within the "File Export" tool. The output was a 3D array (
*#electrodes* : *segment length *: #*trials*). In place of this absent
feature on newer versions of NetStation, our lab was advised by EGI to use
the .mff to .mat file conversion method available through EEGLAB. Here is
the code I've been using to do so (on matlab 2025b)...

"[ALLEEG EEG CURRENTSET ALLCOM] = eeglab;
EEG = pop_mffimport({'data.mff'},{'code'},0,0);
[ALLEEG, EEG, CURRENTSET] = eeg_store( ALLEEG, EEG, 0 );"

Now, when I run this, the ALLEEG.data output contains a 3D array in the
same format as the NetStation 5.4.3-R .mat file  (*#electrodes* : *segment
length *: #*trials*)*. *However, when comparing the same .mff file
converted using the two different methods, there are small discrepancies
across all electrode voltages. The average voltage difference for an
electrode between the EEGLAB and NetStation Tools output is about .001
(this is averaged over the entire segment length and all trials), so I'm
curious to know what might be causing this change between the two methods.
For more context, our sampling rate is 1000 s/s.

Secondly, and potentially related to my first inquiry, I also receive a
string of messages like this in my command window when running the above
code on my .mff file: "Warning: segment discontinuity (1170 samples missing
- pause in the recording or bug?)". I get one of these for each trial (so
1170 in total). I also receive another message that says "eeg_checkset
warning: 3rd dimension size of data (1170) does not match the number of
epochs (1), corrected". It's likely that I'm not understanding some of the
input arguments properly, but I'm not entirely sure what to make of these
messages and if they're related to the discrepancies that I'm seeing. Both
outputs contain the proper amount of electrodes, segment length, and
trials, so I don't see any obvious discrepancies in the 3D array structure
itself. For more context about our processing pipeline, we record all
trials with event codes. We then filter and segment before converting to
.mat.

Dylan Vlasak
-- 
[image: Colgate University email wordmark]
Dylan Vlasak
BRCS Lab Manager
Psychological & Brain Sciences

https://urldefense.com/v3/__http://www.colgate.edu/__;!!Mih3wA!DEMISIZs-dMp1kG4j6Ezt9gnOG_oRR0iv1yO9XNy9w5EU9pYRPKG5gOaNoOpRTFSpc4TgJyTstHl3LkBRggR3PYviQ$ 
he / him / his