I am a relatively new Brainstorm user. I was hoping I could get information regarding what is the best way to perform source estimation from an already preprocessed data from EEGLAB.
I have done all the preprocessing in EEGLAB - downsampling, filtering, bad channel rejection and interpolation, referencing to average reference, bad data rejection using ICA, epoching and baseline correction. I have done these steps on 30 subjects and now I have an average data set of the 30 subjects. This data is stored as a .set file of EEGLAB. The data structure is 256 (channels) X 307 (time points) X 1975 (epochs - combined epochs of all 30 subjects).
I also have the channel location file which I have used in EEGLAB. This is the average channel location file and I have not generated it using a digitizer. It is a 257 channel file, with Cz (channel 257) taken as the reference. I also do not have individual MRIs of any participants.
With this information, it would be really useful if someone could direct me on what is the best way to go about with source estimation.
Welcome to Brainstorm, Anoop!
I encourage you to follow the online tutorials: we believe they explain how to approach the analytical steps along your way. You can also type keywords such as e.g., “MRI template”, “EEG”, “EEGlab” , “Channel template” in the search bar to find sections that address your specific situation.
If you are a new Brainstorm user, we recommend you start by following all the introduction tutorials (section Get started on the Tutorials page), using the example dataset from the tutorial.
Then follow the tutorial "EEG and epilepsy" for additional details about processing EEG.
Once you are familiar with the software, you will have no difficulty importing your processed epochs from an EEGLAB .set file (natively supported by Brainstorm).
- replacing bad channels with local interpolations is not advised before source reconstruction
- import the epochs, not the averaged ERP, and recompute the average in Brainstorm - as you'll see in the tutorial "Noise covariance matrix", you will need an estimate of the noise at the sensor level, and you can't really compute this from an average.