What was new in previous releases?#

Version 0.3 (08/10/2022)#

This new version adds an experimental GUI for annotating ICA components. Please test it out if you find it useful in your workflow! Moreover, we consolidated our testing suite and improved overall coverage and added some bug fixes. We also created a function to extract topomap features from fitted ICA instances.

Enhancements#

Adding a GUI to facilitate the labeling of ICA components by Adam Li and Mathieu Scheltienne (#66)
Improve test coverage and documentation by Mathieu Scheltienne (#94)

Bug#

Ignore the RuntimeWarning issued by the grid inteprolation for ICLabel topographic feature by Mathieu Scheltienne (#69)

API#

Add topographic feature using MNE with get_topomaps by Anand Saini and Mathieu Scheltienne (#71)

Authors#

Version 0.2 (06/30/2022)#

This includes series of bug fixes to the ICLabel ported model. In addition, we have added an experimental feature for helping add ICA component annotations to BIDS-derivative for EEG format.

Enhancements#

Add functions for annotating and labeling ICA components in BIDS format mne_icalabel.annotation.write_components_tsv(), mne_icalabel.annotation.mark_component() by Adam Li (#60)

Bug#

Fix shape of 'y_pred_proba' output from mne_icalabel.label_components() by Mathieu Scheltienne (#36)
Add a warning if the ICA decomposition provided does not match the expected decomposition by ICLabel by Mathieu Scheltienne (#42)
Fix extraction of PSD feature from ICLabel model on epochs by Mathieu Scheltienne (#64)
Fix ICLabel topographic features on ICA fitted with a channel selection performed by picks by Mathieu Scheltienne (#68)

API#

Authors#

Version 0.1 (2022-05-03)#

This is the first version of mne-icalabel that is released. This package consists of a replica of the ICLabel network and pipeline present in EEGLAB for automatically predicting the label of an ICA component and whether it is brain signal, or noise.

We integrated our API on top of MNE-Python and provide a lightweight example demonstrating how to use the ICLabel network.

Authors#

People who contributed to this release (in alphabetical order):