HMM analysis with preprocessed eeglab .set data

[pauldhami]

Dear OHBA Group,

Thank you very much for all the wonderful tools. I have gone over the documentation, but being new to OSL, I still had some outstanding questions.

I have some task data that has been preprocessed with EEGLAB (and is in .set files accordingly).

My goal is to perform a TDE-HMM analysis on this data (in source space).

1. Because I have already preprocessed my data with EEGLAB, can I use the OSL tools to perform only registration and source localization for EEG? Or must I do all my preprocessing with OSL tools?

2. For the EEG dataset I am working with, we do have the individual MRIs, but there were no Polhemus data collected. Can co-registration (and ultimately source localization) be conducted with OSL? How can I manually add the nasion, left, and right fiducials?

Sorry for what may be naive questions, but any insight would be greatly appreciated.

Thank you,
Paul

[woolrich]

Hi Paul, If you are looking to run a TDE-HMM analysis, there is the option available to use the pipeline code associated with this paper: https://pubmed.ncbi.nlm.nih.gov/30210284/ Found here: https://github.com/OHBA-analysis/Quinn2018_TaskHMM. 1. Because I have already preprocessed my data with EEGLAB, can I use the OSL tools to perform only registration and source localization for EEG? Or must I do all my preprocessing with OSL tools? You can do pre-processing elsewhere. However, you do want the data to be in the appropriate SPM12 format at the point of switching over to OSL, so that the co-reg and source recon can work. 1. For the EEG dataset I am working with, we do have the individual MRIs, but there were no Polhemus data collected. Can co-registration (and ultimately source localization) be conducted with OSL? Yes, you should be able to get a (sub-optimal) coreg working in that case. However, you will need to use the SPM12 coreg rather than RHINO. See: https://ohba-analysis.github.io/osl-docs/matlab/osl_example_coregistration.html All the best, Mark. On 18 Feb 2021, at 22:07, pauldhami <notifications@github.com<mailto:notifications@github.com>> wrote: Dear OHBA Group, Thank you very much for all the wonderful tools. I have gone over the documentation, but being new to OSL, I still had some outstanding questions. I have some task data that has been preprocessed with EEGLAB (and is in .set files accordingly). My goal is to perform a TDE-HMM analysis on this data (in source space). 1. Because I have already preprocessed my data with EEGLAB, can I use the OSL tools to perform only registration and source localization for EEG? Or must I do all my preprocessing with OSL tools? 2. For the EEG dataset I am working with, we do have the individual MRIs, but there were no Polhemus data collected. Can co-registration (and ultimately source localization) be conducted with OSL? Sorry for what may be naive questions, but any insight would be greatly appreciated. Thank you, Paul — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub<#68>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AALVITPRY3FEUZI3AJ37DOTS7WFT5ANCNFSM4X3GC6GA>.

[pauldhami]

Dear Dr. Woolrich,

thank you very much for your help, it has got me up and running.

I am following the this tutorial to see if I can at least first do a simple sensor contrast with one condition (my epoched EEG data has just one condition).

https://ohba-analysis.github.io/osl-docs/matlab/osl_example_oat_1_sensorspace.html

Before following the example, I converted my eeglab .set file to the appropriate SPM M/EEG epoched format.

I then followed each step, but am running into the following error, which even after going through the code, I am unsure as to why it is occurring.

>> oat = osl_run_oat(oat);
Detected EEG data. Using default EEG settings
Warning: oat.source_recon.modalities not set, or not set properly. Will set to default:EEG 
> In osl_check_oat (line 107)
  In osl_run_oat (line 13) 
oat.source_recon.D_epoched set. OAT will do an epoched data trial-wise GLM
*************************************************************
Running source_recon
*************************************************************
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%  RUNNING OAT SOURCE RECON (SENSOR SPACE SETUP) ON SESS = 1  %%%%%%%%%%%%%%%%%%%%%%%
Using epoched data as input
Preparing source recon stage for /Users/prabhjotdhami/Desktop/HMM/OSL_Source_Preprocessing/yd-c001/spm_files/spmeeg_yd-c001_v3_lmc_cleaned_interp.mat
Will be designated session1

SPM12: spm_eeg_copy (v5079)                        20:27:00 - 22/02/2021
========================================================================

SPM12: spm_eeg_copy (v5079)                        20:27:01 - 22/02/2021
========================================================================
No temporal filtering applied \n\n
Source recon OAT stage has failed

ans =

    'Unrecognized function or variable 'D_epoched'.
     
     Error in oat_prepare_source_recon (line 244)
         S.D=D_epoched;
     
     Error in oat_run_source_recon_sensorspace (line 63)
         [source_recon_sess source_recon_sess_results source_recon_sess_results.report] = oat_prepare_source_recon(source_recon_sess, report_sess);
     
     Error in osl_run_oat (line 51)
                 [results_fnames results]=oat_run_source_recon_sensorspace(oat);'

By not specifying oat.source_recon.normalise_method = 'none', it seems to run okay, but I am not sure what I had to comment this line out for it to run when in the tutorial, it runs fine.

Also, if I may also ask, I am following the OAT 3 source analysis tutorial (https://ohba-analysis.github.io/osl-docs/matlab/osl_example_oat_3_sourcespace.html)

Although I am using mainly default input parameters for EEG, I run into this error when running osl_run_oat(oat):

    'Error using osl_inverse_model (line 63)
     The value of 'pca_order' is invalid. It must satisfy the function: @(x)isnumeric(x)&&isscalar(x)&&x>0&&~mod(x,1).
     
     Error in oat_run_source_recon (line 165)
         D = osl_inverse_model(D,mni_coords,S2);
     
     Error in osl_run_oat (line 53)
                 [results_fnames results]=oat_run_source_recon(oat);
     
     Error in LiveEditorEvaluationHelperESectionEval1c5e202b (line 157)
     oat = osl_run_oat(oat);
     
     Error in matlab.internal.editor.evaluateCode'

Any suggestions as to what I may be doing wrong would be great.

Also, apologizes if I should not be posting such questions here. Would the SPM mailing list be better?

Thank you in advance,
Paul

[pauldhami]

Apologizes to add to this thread, but I think I figured out my past issue in regards by specifying oat.source_recon.pca_dim before calling osl_run_oat(oat).

However, I am now running into the following error, which I am not having much luck troubleshooting.

Item 'Jacobian modulation?', field 'val': No matching defaults value found.
Item 'Unwrapping method', field 'val': No matching defaults value found.
Item 'FWHM', field 'val': No matching defaults value found.
Item 'pad', field 'val': No matching defaults value found.
Item 'Weighted smoothing', field 'val': No matching defaults value found.
Item 'Template image for brain masking', field 'val': No matching defaults value found.
Item 'FWHM', field 'val': No matching defaults value found.
Item 'Number of erosions', field 'val': No matching defaults value found.
Item 'Number of dilations', field 'val': No matching defaults value found.
Item 'Threshold', field 'val': No matching defaults value found.
Item 'Regularization', field 'val': No matching defaults value found.
Item 'Defaults File', field 'val': No matching defaults value found.
Item 'VDM filename extension', field 'val': No matching defaults value found.
Item 'Jacobian modulation?', field 'val': No matching defaults value found.
Item 'Unwrapping method', field 'val': No matching defaults value found.
Item 'FWHM', field 'val': No matching defaults value found.
Item 'pad', field 'val': No matching defaults value found.
Item 'Weighted smoothing', field 'val': No matching defaults value found.
Item 'Template image for brain masking', field 'val': No matching defaults value found.
Item 'FWHM', field 'val': No matching defaults value found.
Item 'Number of erosions', field 'val': No matching defaults value found.
Item 'Number of dilations', field 'val': No matching defaults value found.
Item 'Threshold', field 'val': No matching defaults value found.
Item 'Regularization', field 'val': No matching defaults value found.
Item 'Defaults File', field 'val': No matching defaults value found.
Item 'VDM filename extension', field 'val': No matching defaults value found.
Item 'Distortion direction', field 'val': No matching defaults value found.
Item tools: No field(s) named
beamforming
Item tools: No field(s) named
beamforming
Item tools: No field(s) named
beamforming
Item tools: No field(s) named
beamforming
Item tools: No field(s) named
beamforming
Item tools: No field(s) named
beamforming
Source recon OAT stage has failed

ans =

    'Error using spm_jobman>fill_run_job (line 461)
     No executable modules, but still unresolved dependencies or incomplete module inputs.
     
     Error in spm_jobman (line 247)
             sts = fill_run_job('run', cjob, varargin{3:end});
     
     Error in osl_inverse_model (line 244)
     spm_jobman('run',matlabbatch)
     
     Error in oat_run_source_recon (line 165)
         D = osl_inverse_model(D,mni_coords,S2);
     
     Error in osl_run_oat (line 53)
                 [results_fnames results]=oat_run_source_recon(oat);'

Any help with regards to figuring this out would be greatly appreciated.

Thank you in advance,
Paul