search

lina-usc / pylossless

🧠 EEG Processing pipeline that annotates continuous data
https://pylossless.readthedocs.io/en/latest/
MIT License
24 stars 8 forks source link

Fix the montage rereference #131

Open christian-oreilly opened 1 year ago

christian-oreilly commented 1 year ago

Currently the pipeline function wrap_locs appears not be used anywhere. Similarly, self.config['replace_string']['montage_info'] is used only in wrap_locs. I also think we did not implemented the referencing on a common surface. We need to fix these things.

scott-huberty commented 1 year ago

self.config['replace_string'] definitely sounds like a hold over from the MATLAB config, If I recall correctly.

I also think we did not implemented the referencing on a common surface.

True. But we never decided on whether it's worth the effort?

christian-oreilly commented 1 year ago

True. But we never decided on whether it's worth the effort?

Indeed. That is the point... we should decide and act on it (i.e., remove wrap_locs if not, implement it if yes). Following the previous exchange with @jadesjardins I tend to lean on the side that it is relevant to do something... particularly for the pooling of dataset recording with different systems and having a different coverage of the head (e.g., EGI including the lower back of the head vs a 10-20 montage that did not) since this coverage difference can bias the reference.

self.config['replace_string'] definitely sounds like a hold over from the MATLAB config, If I recall correctly.

Yes, but I am not so much reference to the 'replace_string' of that code but the 'montage_info' and the fact that we currently have two place where the montage can be specified.

scott-huberty commented 1 year ago

Indeed. That is the point... we should decide and act on it (i.e., remove wrap_locs if not, implement it if yes). Following the previous exchange with @jadesjardins I tend to lean on the side that it is relevant to do something...

Agreed. Now we just have to find someone to do the job ;)

Andesha commented 1 year ago

I'm happy to take a stab at something like this. I'm about to push a minor montage fix so it's good timing.

Can you point me to the MNE functions that would do something like warp_locs to a standard surface sort of thing?

scott-huberty commented 1 year ago

@Andesha I haven't looked at dyour comment/PR Yet but yeah the co-registration to the 10-20 surface (to be used for robust average referencing) is something we want to implement but it's a bit complex so never finished it.

If you look at the warp_locs code you'll see what mne function is called, and you can check the imports to see what module it is coming from.

The MNE site will definitely have an API documentation and probably a tutorial for it.

Ping us with any questions, we're happy to help.