[FieldTrip] Problems with ft_Channelrepair / interpolation

[Schoffelen, J.M. (Jan Mathijs)]

Hi Kinkini,

I don’t have any concrete suggestions here. Perhaps you should try and first pinpoint the potential cause of the issue yourself. If you dump an entire pipeline in an e-mail, it is unclear to us where we should focus on. Specifically, it is not clear whether any upstream processing steps are relevant at all for the issue you are facing. You wrote: "but when I wanted to apply statistics on the data, I had errors”.  This is so ill-conditioned that we can just guess as to what is going wrong for you.
Perhaps consult http://www.fieldtriptoolbox.org/faq/how_to_ask_good_questions_to_the_community/ to phrase your question a bit more to-the-point.

Good luck,

Jan-Mathijs


On 2 May 2020, at 22:22, Kinkini . <bkinkini at gmail.com<mailto:bkinkini at gmail.com>> wrote:

Hello  Jan,

Thank you so much for your comments.

I am excluding bad channels while applying CAR.
Also, I tried excluding bad channels from ICA and interpolate them later. But as I had mentioned before, I am facing problems applying ft_freqstatistics.

Any suggestion on that?

Thanks again,

Kinkini


On Sat, May 2, 2020 at 1:51 PM Schoffelen, J.M. (Jan Mathijs) <jan.schoffelen at donders.ru.nl<mailto:jan.schoffelen at donders.ru.nl>> wrote:
Hi Kinkini,

I would recommend not doing a CAR before you have removed bad channels.
Also, I would recommend not tu ‘repair’ any bad channels prior to running ICA.
Best wishes,
Jan-Mathijs


On 1 May 2020, at 13:33, Kinkini . <bkinkini at gmail.com<mailto:bkinkini at gmail.com>> wrote:

Dear FieldTripers,

Hope everyone is coping well with the current situation.

I am relatively new in fieldtrip and I have been long stuck with using ft_channelrepair for quite sometime now.

Briefly about my analysis pipeline:
1- Loading and epoching the data
2- ft_preprocessing - filtering, baselining, CAR
3- ft_databrowser - marking bad channels
4. ft_rejectvisual - rejecting bad channels; keepchannel = 'repair'
5. ft_channelrepair
6. ICA
7. TF Analysis
8. TF statististics

I have problems pertaining to step 4 and 5. In step 4 I want to keep the bad channels so that I can interpolate them in the next step. When I use data browser to look thorough the data, the bad channels look okay. But the problem starts when I perform ICA on them. My first observation is that the wchange value is very big (looks like this : 257868.69202720), but ICA runs completely. However the ICA component topoplots look weird (screenshots attached). However, when I look through the icabrowser plots , there are peaks at physiological frequencies.

In order to avoid this, I tried getting rid of the bad channels completely using ft_rejectvisual, perform ICA and then get back the missing channels using neighbourhood average. Doing this, I was able to perform TF analysis, but when I wanted to apply statistics on the data, I had errors. Finally, I had to avoid the entire step of  interpolation to be able to run everything and also perform statistics. But in a long run, I cannot avoid this step.

I am finding no clue as to where is it going wrong.

Hence, any idea, any clue from you all would help me very much. I am happy to provide with more details if necessary.
Thank you so much for your response.

Best regards,

Kinkini Bhadra
PhD student
Department of Fundamental Neurosciences
University of Geneva


<Screenshot 2020-05-01 11.57.28.png><Screenshot 2020-05-01 11.57.42.png>_______________________________________________
fieldtrip mailing list
https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
https://doi.org/10.1371/journal.pcbi.1002202

_______________________________________________
fieldtrip mailing list
https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
https://doi.org/10.1371/journal.pcbi.1002202
_______________________________________________
fieldtrip mailing list
https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
https://doi.org/10.1371/journal.pcbi.1002202

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20200503/9f777960/attachment.htm>