Importing Resting State Data

Hello Brainstorm Community,

I would like to import my resting state MEG recordings (3 minutes = 180s) after cleaning my data using the linked file system presented in the tutorials. However, upon importing, BS asks to select my epochs (again, resting state, so no events). In the past, I've used 120 seconds and divvied them into 2 second epochs (for a total of 60 raw files). I cannot compute the source localisation on the raw files because they're not continuous, and as such I have to average them. The result is a 2 second file which is the average of all 60 epochs per subject.

However, I have 2 problems with this method:

  1. After projecting to the default anatomy, and trying to compute the PSD, I get a warning that the 2 seconds aren't enough. In other words, the 2 second data form is limiting analyses.

  2. Averaging, to me, means I lose a lot of data. Condensing entire recordings to 2 seconds seems anti-climatic.

So, my question is as follows:

Is there a way to compute sources on my entire recordings without splitting the data into epochs when I import them to the database? My idea is to create one epoch that defines the length of my recordings (epoch length: 0-180s, one epoch). If that is not advisable, what would be a recommended epoch length?

The resting OMEGA tutorial uses the CTF system which lets them convert their raw recordings (without the need to average) between epochs and continuous data, a luxury I don't have. I get the error message "Conversion from epoched to continuous is only available for CTF .ds files" when I try to do that.

My warmest gratitude for your time and response.

Best,

Aquila.

Hello,

In the past, I’ve used 120 seconds and divvied them into 2 second epochs (for a total of 60 raw files).

Why did you split the data in 2s blocks in the the first place?
You could use the menu “Review raw file” instead of “Import MEG/EEG”:
http://neuroimage.usc.edu/brainstorm/Tutorials/ChannelFile#Review_vs_Import

I cannot compute the source localisation on the raw files because they’re not continuous,

You do not need any data to compute the inverse model. Right-click on the head model > Compute sources [2016].
http://neuroimage.usc.edu/brainstorm/Tutorials/SourceEstimation#Computing_sources_for_single_trials

as such I have to average them

Averaging blocks of 2s resting recordings together is not indicated for any kind of analysis. Averaging epochs is only useful when the files are time-locked on an event of interest, to increase the signal-to-noise ratio of the brain signals that are related with this event. You need to compute a measure from your recordings before averaging your files, for instance power or coherence measures.

After projecting to the default anatomy, and trying to compute the PSD, I get a warning that the 2 seconds aren’t enough. In other words, the 2 second data form is limiting analyses.

To compute the PSD of continuous resting recordings, follow the procedure presented in this tutorial:
http://neuroimage.usc.edu/brainstorm/Tutorials/RestingOmega#Power_maps
If you do not have access to the original files: select all the epochs in the Process1 box, run the process “Standardize > Uniform epoch time” followed by “Frequency > Fourier transform”, with the option “Save average power”.

Cheers,
Francois