[Eeglablist] Applying ICA Weights to differently filtered data

[Makoto Miyakoshi]

Hi Naviya,

The ICs look ok to me.

In some of them, I see DC fluctuation in the ERP image (bottom left). This
is due to using 0.1-Hz high-pass filter and without 'baseline correction'
after epoching? It's not a problem, but if you care, applying post-ICA
'baseline correction' is fine.

Makoto

On Mon, Oct 19, 2026 at 2:11 PM Naviya Lall via eeglablist <
eeglablist at sccn.ucsd.edu> wrote:

> Hi all,
>
> I have a question about the components that form after we have applied ICA
> weights from a high pass filtered dataset (1-50hz) to a differently
> filtered dataset of 0.1-50hz.
>
> Here is the link to what the components look like-
>
> https://urldefense.com/v3/__https://drive.google.com/drive/folders/1_uHIGHP12CthxxWR-0NNcpw_yCf0IbrI__;!!Mih3wA!A5LzkbksQlXQD7fl3T5zm__ojTThROP8DzwosEyUxcBJmTNrqi8llu0kbExMPXNXG0fDAeOxMWu0XX11PWv7wFPp$
>
> My question is- *are the components supposed to look like this after
> applying the weights? *Particularly the Epoched IC activity which shows
> trial wise data (under the topo plot) plot. Some components like the eye
> components show up well. Others seem to be odd looking. We plan to perform
> ERP analysis after this which is why I am looking for confirmation on this.
>
> These are the steps used-
> Preprocessing steps followed:
>
>    1. Import raw file 32 channels EEG data xdf file
>    2. Downsample data 500 to 250hz
>    3. Apply notch filter 48-52Hz
>    4. Bandpass filter from 1-50Hz
>    5. Add channel location + remove marker channel (data was recorded on
>    LSL which had a marker channel with no data)
>    6. Epoch from -0.3 to 3s (I have tested running ICA with continuous data
>    as well but it was noisier)
>    7. Re-reference to average
>    8. ICA default
>    9. New dataset of the same session i.e. import the raw file again:
>
>    a. Downsample 500 to 250
>    b. Notch filter 48-52hz (same as before)
>    c. bandpass from 0.1 Hz to 50Hz
>    d. Add channel location + remove marker channel
>    e. Epoch from -0.3 to 3s
>    f. re-referenced to average
>    g. For ICA: Edit>Dataset info> import ICA from another dataset (where
> ICA was applied to 1-50hz filtered file i.e. point 8.  referred to these
> steps by Makoto Miyakoshi-
>
> https://eeglab.ucsd.edu/wiki/Makoto's_preprocessing_pipeline#High-pass_filter_the_data_at_1-Hz_(for_ICA,_ASR,_and_CleanLine)(05/18/2022_updated) 
>
>    h. Label components of eye, muscle, channel noise
>    i. remove components and save the file
>
> Thank you!
>
>
> Best regards,
> Naviya
>
> --
> Naviya Lall
> Junior Research Fellow
> Cognitive Science Lab
> IIIT Delhi
> naviyalalluni.wixsite.com <
> https://urldefense.com/v3/__https://naviyalalluni.wixsite.com/naviyalall__;!!Mih3wA!A5LzkbksQlXQD7fl3T5zm__ojTThROP8DzwosEyUxcBJmTNrqi8llu0kbExMPXNXG0fDAeOxMWu0XX11PbsQM5lg$
> >
> _______________________________________________
> To unsubscribe, send an empty email to
> eeglablist-unsubscribe at sccn.ucsd.edu or visit
> https://sccn.ucsd.edu/mailman/listinfo/eeglablist  .
>