# Automatically annotate bad data segments

**URL:** <https://mne.discourse.group/t/automatically-annotate-bad-data-segments/3697>\
**Category:** Support & Discussions\
**Tags:** preprocessing, eeg\
**Created:** [September 15, 2021, 9:17am UTC](https://mne.discourse.group/t/automatically-annotate-bad-data-segments/3697 "2021-09-15T09:17:26Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![mscheltienne](https://yyz2.discourse-cdn.com/free1/user_avatar/mne.discourse.group/mscheltienne/32/827_2.png) [@mscheltienne](https://mne.discourse.group/u/mscheltienne)\
**Post date:** [September 15, 2021, 9:17am UTC](https://mne.discourse.group/t/automatically-annotate-bad-data-segments/3697/1 "2021-09-15T09:17:26Z")

</div>

One feature I really enjoy about MNE is the possibility to annotate interactively bad data segments on the plot interface. I’m usually not specifying the source of the noise, and I use a generic `BAD` label when I see a very noisy segment.

However, for large volumes of data, manual annotation becomes cumbersome. Is there an MNE method, or a library in the MNE ecosystem that automatically annotates bad segments of data?

I’m looking for an equivalent to `pyprep` which is for bad (noisy) channels detection.

---

<div class="post-metadata">

**Author:** ![agramfort](https://avatars.discourse-cdn.com/v4/letter/a/7c8e57/32.png) [@agramfort](https://mne.discourse.group/u/agramfort)\
**Post date:** [September 15, 2021, 9:48am UTC](https://mne.discourse.group/t/automatically-annotate-bad-data-segments/3697/2 "2021-09-15T09:48:51Z")

</div>

training ML models to do this has been my dream for years but I need enough  
curated and well organized data…

Alex

---

<div class="post-metadata">

**Author:** ![mscheltienne](https://yyz2.discourse-cdn.com/free1/user_avatar/mne.discourse.group/mscheltienne/32/827_2.png) [@mscheltienne](https://mne.discourse.group/u/mscheltienne)\
**Post date:** [September 15, 2021, 4:53pm UTC](https://mne.discourse.group/t/automatically-annotate-bad-data-segments/3697/3 "2021-09-15T16:53:48Z")

</div>

One day… one day…  
Aside from advanced ML models, is there some pipeline, algorithm which do this with simple techniques?  
Basically, as long as I don’t have to code it from scratch, it’s worth integrating into my pipeline.

---

<div class="post-metadata">

**Author:** ![agramfort](https://avatars.discourse-cdn.com/v4/letter/a/7c8e57/32.png) [@agramfort](https://mne.discourse.group/u/agramfort)\
**Post date:** [September 16, 2021, 7:55am UTC](https://mne.discourse.group/t/automatically-annotate-bad-data-segments/3697/4 "2021-09-16T07:55:12Z")

</div>

I am not aware of such tool

A