# Get source type probabilities from ica\_label

**URL:** https://mne.discourse.group/t/get-source-type-probabilities-from-ica-label/7356
**Category:** Support & Discussions
**Tags:** preprocessing, eeg, ica
**Created:** [August 10, 2023, 11:41am UTC](https://mne.discourse.group/t/get-source-type-probabilities-from-ica-label/7356 "2023-08-10T11:41:51Z")
**Posts on this page:** 1
**Page:** 1

<div class="post-metadata">

### Author: ![jasonh](https://avatars.discourse-cdn.com/v4/letter/j/e0b2c6/32.png) [@jasonh](https://mne.discourse.group/u/jasonh)
#### Post date: [August 10, 2023, 11:41am UTC](https://mne.discourse.group/t/get-source-type-probabilities-from-ica-label/7356/1 "2023-08-10T11:41:51Z")

</div>

MNE 1.4.2 on macOS 13.4.1

I’m using the `mne-icalabel` package to classify ICA components automatically. Using `mne_icalabel.label_components` I can get a label for each component together with the probability of the classification.

Does anyone know a way to also retrieve the other probabilities of the less likely source types from mne-icalabel? Like in this ICLabel implementation in EEGLab: [The EEGLAB News](https://sccn.ucsd.edu/eeglab/eeglab_news/3/ICLabel_Q&A.php)

E.g., I’m currently getting an IC labelled as “muscle” with p = 0.87, but I would also like to know where the remaining .13 are going, so I can limit the exclusion of components to those where “brain”/“other” is below 0.05.