Contributing a preprocessing method & dataset?

Support & Discussions
,
1

We are currently working to include an adaptation of the PCA-OBS algorithm (https://www.iannettilab.net/uploads/1/2/3/5/123514110/removal_of_fmri_artifacts.pdf) to mne-python/mne/preprocessing on Github.

This algorithm is typically used to remove the ballistocardiographic artefact from simultaneous EEG-fMRI, but we have adapted it by removing the time delay such that it operates to remove the cardiac artefact in EEG, but especially ESG (electrospinography) data. In order to create a good example showcasing the method, we would also like to contribute a combined EEG&ESG dataset - is this possible?

Although MNE-python is typically used by brain researchers, there is a growing community also interested in incorporating spinal recordings, for which this method is particularly useful :slight_smile:

2

Hello,

Yes it is possible. You could add it as a downloadable dataset to the mne.datasets module. You’ll need to:

Mathieu

3

Thank you for the reply! The datasets are already openly available via OpenNeuro, would this count as somewhere ‘accessible’?

4

@Bailey Happy Cake Day!! :cake:

5

Yes and no. You could use openneuro-py within your example/tutorial to fetch the dataset, but datasets within the mne.datasets module are archived while on openneuro they are stored file-by-file as git-like repository (I think, might be mistaken). I don’t believe the pooch-based dataset fetcher in mne.datasets can handle openneuro datasets, maybe @richard and @adam2392 might know more.

Mathieu

6

Ok cool, I’ll look at how to upload the actual dataset but good to know I can just use openneuro-py as a backup!

7

Hahahaha thank you - 3 long years of asking questions and now I might actually contribute something eh!

1 Like
8

I don’t think mne.datasets easily integrates with opeknenuro yet but I could be wrong. Haven’t touched that submodule in awhile.

maybe it should given the datasets available online. Perhaps open a GH issue to discuss?