[Eeglablist] Study ICA Clustering Error

Flanagan, Shawn D sdf29 at pitt.edu
Sat Dec 23 08:30:44 PST 2023


Hello,

I am working through the EEGlab Group Analysis Pipeline and am running into the following error when I attempt to execute the Study --> PCA Clustering (original) --> Build preclustering array step: "Cannot process components from different ICA decompositions of the same subjects". See below.

[image.png]

My participants have two sessions (different days/cap placements) and within each session, four different trial types (2 stimulation targets and 2 conditions for each target). The preprocessing steps (including ICA runs) were run on the complete dataset for each session (all four trial types), with session data analyzed separately as recommended.

I am not sure where to begin with this - I am working with TEP data and used the TESA plug-in / pipeline with an initial round of fastICA to remove stimulation-induced artifacts. and a second round of fastICA to remove all other artifacts. I saved the component weights at both steps. After I received this error the first time, I re-ran the analysis beginning at the second ICA decomposition using EEGlab's group analysis batching pipeline (infomax 1 extended --> ICAlabel --> ICAlabel artifacts --> DIPfit head model and settings --> DIPFIT locate dipoles (auto). I then precomputed the component measures and attempted to build the preclusting array (PCA - I can provide parameters if potentially consequential but not my impression at this point). Same issue. I then adjusted the study design - originally, I set up two designs - with categorical session (1 2)  and categorical condition (con stim) as independent variables for each target (M1 PFC). When this didn't work, I reorganized the study design to have one study with 2 sessions (1 2) and 4 conditions (M1con M1stim PFCcon PFCstim). No luck. I should note that the procedure seems to work when the preclustering array is programmed solely based on scalp maps, but of course, nothing other than scalp maps will show in the view and edit current component clusters pup-up menu (same error displayed when e.g.,plot ERPs, plot spectra, etc).

Appreciate your time and guidance! Please let me know if any other information would be helpful.

All the Best,

Shawn


---
Shawn Flanagan, Ph.D.
Associate Professor

Department of Sports Medicine

McGowan Institute of Regenerative Medicine

Director, Sports Medicine PhD Program

School of Health and Rehabilitation Sciences

University of Pittsburgh

3860 South Water Street
Pittsburgh, PA 15203
sdf29 at pitt.edu<mailto:sdf29 at pitt.edu> Email
412.246.0460 Office
630.854.0547 Mobile

Neuromuscular Research Laboratory<https://urldefense.com/v3/__http://www.nmrl.pitt.edu/__;!!Mih3wA!BcxIORCs0LPbB9-s5srvA3gyaXr_RuVP8RpmbFClGW33wWEVLneUT8dRc9o6GvhQRFrrlU_HhTidGtmcrODb$ >




More information about the eeglablist mailing list