# Visually picking out bad channels

**URL:** <https://mne.discourse.group/t/visually-picking-out-bad-channels/7458>\
**Category:** Support & Discussions\
**Created:** [August 29, 2023, 11:38pm UTC](https://mne.discourse.group/t/visually-picking-out-bad-channels/7458 "2023-08-29T23:38:44Z")\
**Posts on this page:** 5\
**Page:** 1

<div class="post-metadata">

**Author:** ![binary\_bits](https://avatars.discourse-cdn.com/v4/letter/b/9dc877/32.png) [@binary\_bits](https://mne.discourse.group/u/binary_bits)\
**Post date:** [August 29, 2023, 11:38pm UTC](https://mne.discourse.group/t/visually-picking-out-bad-channels/7458/1 "2023-08-29T23:38:44Z")

</div>

**❓ If you have a question or issue with MNE-Python, please include the following info:**

- MNE version: 3.10
- operating system: macOS Ventura 13.0

Hi everybody,

I am learning to visually pick out bad channels in raw MEG data. I am following the tutorial here [Handling bad channels — MNE 1.5.0 documentation](https://mne.tools/stable/auto_tutorials/preprocessing/15_handling_bad_channels.html).

I used raw.plot() to look at the raw MEG data. Here is what I have (screenshot attached). This looks completely different from the example data from the MNE website. I see large areas of blue in my data. Is this normal? I am new to this so would appreciate any help/advice on what I am looking at.

 ![sub-01](https://global.discourse-cdn.com/free1/uploads/mne/original/2X/c/c6249ec028aa46fd3f5b10c4ff6313a1cf813812.png)

Thank you for your help!

---

<div class="post-metadata">

**Author:** ![jstout211](https://yyz2.discourse-cdn.com/free1/user_avatar/mne.discourse.group/jstout211/32/180_2.png) [@jstout211](https://mne.discourse.group/u/jstout211)\
**Post date:** [August 30, 2023, 12:31am UTC](https://mne.discourse.group/t/visually-picking-out-bad-channels/7458/2 "2023-08-30T00:31:57Z")

</div>

If you hit the minus key a bunch of times, it will bring your sensors into range.

You may have already done this, but notch\_filtering (raw.notch\_filter([60,120,180])) the raw data and possibly t/SSS-ing your data (again you may have already done this) will clean up the signal a little bit.

–Jeff

---

<div class="post-metadata">

**Author:** ![binary\_bits](https://avatars.discourse-cdn.com/v4/letter/b/9dc877/32.png) [@binary\_bits](https://mne.discourse.group/u/binary_bits)\
**Post date:** [August 30, 2023, 5:02pm UTC](https://mne.discourse.group/t/visually-picking-out-bad-channels/7458/3 "2023-08-30T17:02:01Z")

</div>

Hello,

Thank you so much for your reply.  
I was under the impression that you pick out the bad channels first before running the maxwell\_filter/tsss-ing raw data.

Which method is better?

(1) tsss raw MEG data → manually pick out bad channels → run tsss again on data with the bad channels omitted

(2) low pass filter (55Hz) → manually pickout bad channels → run tsss

Or do both methods accomplish the same thing?

I understand that it is important to pick out the bad channels early or else I will have to redo everything.

Thanks again for your help.

---

<div class="post-metadata">

**Author:** ![jstout211](https://yyz2.discourse-cdn.com/free1/user_avatar/mne.discourse.group/jstout211/32/180_2.png) [@jstout211](https://mne.discourse.group/u/jstout211)\
**Post date:** [August 30, 2023, 5:26pm UTC](https://mne.discourse.group/t/visually-picking-out-bad-channels/7458/4 "2023-08-30T17:26:49Z")

</div>

Hi @binary_bits ,

I forgot about the picking bad channels first before doing t/SSS - but you are correct, you should pick the bad channels first and then do the t/SSS process - or else it will corrupt your SSS-ed data. (I haven’t worked in an Elekta/MEGIN site in several years).

There is an automated bad\_channel/flat\_channel picker that uses the maxwell processing – [mne.preprocessing.find\_bad\_channels\_maxwell — MNE 1.6.0.dev29+gfe9358f16 documentation](https://mne.tools/dev/generated/mne.preprocessing.find_bad_channels_maxwell.html#mne.preprocessing.find_bad_channels_maxwell)

And regarding choice 1 or 2 – I wouldn’t even filter your data - just look at the raw in the correct amplitude and you will see popping channels / jumps or very high freq noise.

Also - just an FYI I think that the plot command has a filtering option, so that you can just filter for visualization but not actually filter your data. Just look at the options in the raw.plot? . I think there you can also set a standard amplitude different from the default so that you dont need to reduce it after loading.

–Jeff

---

<div class="post-metadata">

**Author:** ![binary\_bits](https://avatars.discourse-cdn.com/v4/letter/b/9dc877/32.png) [@binary\_bits](https://mne.discourse.group/u/binary_bits)\
**Post date:** [August 30, 2023, 10:31pm UTC](https://mne.discourse.group/t/visually-picking-out-bad-channels/7458/5 "2023-08-30T22:31:09Z")

</div>

thank you so much!