Hi Francois,
I'm exporting subject ERP averages from an extracted single electrode 'cluster' Brainstorm into Matlab for analysis purposes. However, I noticed that the grand average of the subject averages in the exported cluster is actually somewhat different from the grand average of the same subject averages displayed in Brainstorm (blue trace). Attached is a comparison of the two grand averages...
Any ideas what might be the cause? I've spent hours trying to find something wrong in my analysis... I have the latest version of Brainstorm and I'm running it on Windows 7 with Matlab R2012b. As always, thanks a lot for your help!
-Avi
Hi Avi,
You are comparing the two approaches:
- Select all the trials > Process “extract cluster time series” > Process “average rows”
- Select the average of the trials > Process “extract cluster time series”
You should indeed get exactly the same results (except for rounding errors < 1e-13), the two operations should be equivalent.
This is EEG? Do you have bad channels and/or bad trials in the list? Are you using an average reference?
Those differences could come from different ways of calculating the average or the reference in the two cases.
The average process takes into account the bad trials and bad channels, but not the clusters-related processes.
To test if this has an influence, you could make a copy of your subject, mark all the channels and trials as good, and try again.
Stop by my office with your data if you need additional help with this.
Francois
Hello Francois,
My name is Nithin and i am a post doc at University of MD. I am working on an experiment using continuous match the tone in controls and schizophrenia patients. I am a physician researcher, hence no experience with computation or matlab scripting, however i do curb side sometimes with a programmer colleague. I am planning to examine two condition (correct versus incorrect response) between patients and controls, i do have structural images but not able to use the nif files after converting from Afni format and my mentor says our images do not have details to register the 3 coordinates. So i decided to use the default images and started the tutorial, however i am using .eeg files from neuroscan that contains all the epochs embedded (correct and incorrect), not sure how to phrase the conditions . How would you suggest i set my experiment up…should i sort the correct incorrect at the neuroscan level and save separately, do i average the conditions in brainstorm or should i average this from neuroscan?
I am more interested in time-frequency analysis: Inter-trial phase coherence, Transfer entropy and Granger causality. Today i tried to run the time frequency analysis but i was not able to run the analysis kept having error messages (I was trying on one subject). Any help would be deeply appreciated; if there is a possibility of teleconferencing I would be delighted to discuss this further.
Hi Nithin,
How would you suggest i set my experiment up...should i sort the correct incorrect at the neuroscan level and save separately, do i average the conditions in brainstorm or should i average this from neuroscan?
The best possible option is for you to import the continuous Neuroscan file (.cnt) instead of the epoched file (.eeg) then to epoch and average in Brainstorm, as illustrated in the introduction tutorials on continuous recordings.
Inter-trial phase coherence, Transfer entropy and Granger causality
No current implementation in Brainstorm for the two first. The Granger causality is there but still at a very early stage of development.
If you are not very familiar with signal processing and programming, I would recommend you use only the tools that are correctly documented on the Tutorials page (http://neuroimage.usc.edu/brainstorm/Tutorials).
We will preparer a tutorial for the functional connectivity features later this year.
Today i tried to run the time frequency analysis but i was not able to run the analysis kept having error messages
Please copy-paste the full error message and describe what type of operation you were trying to run, and the type of input data.
Cheers,
Francois