Back to PLS HelpPLS on EEG Data Freq Bins
Posted on 06/25/09 20:13:35
Number of posts: 41
Hi all ~
My question is similar to Jim's ('other' data PLS) but regarding an EEG analysis on frequency bins. My datamat will be [subjects x conditions x groups] in rows and [electrodes x freq bins] in columns. Essentially I am replacing the time window in a typical EEG analysis, with a window of 1 Hz frequency bins.
I assume the simplest way to do this analysis is to use the command line pls_analysis.m script. However, I wonder if I could benefit in any way from the PLS GUI analysis because it allows for visualization of the data. It would allow me to view the results across all electrodes at once in a visual display - this is helpful, and EEG data 'viewers' are not readily available (as far as I know - perhaps EEGLAB can be of use here?).
1. My first question is - is this assumption about visualization correct, or am I wasting my time and should stick to the command line?
Assuming I were to use the GUI - my concern is that there is a default baseline correction in setting up the ERP datamat. Clearly I do not want my any baseline correction applied to a window of requency bins. I figure that I could add in a fake bin into my frequency window - with the value zero in it. I assume this will ensure that the correction will not change the data.
2. So my second question is, is this thinking correct (i.e., a 0 value 'baseline' will result in a correction that doesn't change the data)?
Many thanks!
Agatha
My question is similar to Jim's ('other' data PLS) but regarding an EEG analysis on frequency bins. My datamat will be [subjects x conditions x groups] in rows and [electrodes x freq bins] in columns. Essentially I am replacing the time window in a typical EEG analysis, with a window of 1 Hz frequency bins.
I assume the simplest way to do this analysis is to use the command line pls_analysis.m script. However, I wonder if I could benefit in any way from the PLS GUI analysis because it allows for visualization of the data. It would allow me to view the results across all electrodes at once in a visual display - this is helpful, and EEG data 'viewers' are not readily available (as far as I know - perhaps EEGLAB can be of use here?).
1. My first question is - is this assumption about visualization correct, or am I wasting my time and should stick to the command line?
Assuming I were to use the GUI - my concern is that there is a default baseline correction in setting up the ERP datamat. Clearly I do not want my any baseline correction applied to a window of requency bins. I figure that I could add in a fake bin into my frequency window - with the value zero in it. I assume this will ensure that the correction will not change the data.
2. So my second question is, is this thinking correct (i.e., a 0 value 'baseline' will result in a correction that doesn't change the data)?
Many thanks!
Agatha
Replies:
Untitled Post
I'm Online
Posted on 06/26/09 02:10:29
Number of posts: 229
Agatha..
you are correct ..
you can use the gui for analysis and display of the histogram data per channel.
you will need to 'calculate' an appropriate 'sampling rate' to match the number of bins, and set the prestim baseline to zero
this should allow you to load the data in from a text file as if it were ERP data
we do not do any 'zeroing' of the ERP data with respect to onset/baseline.
nancy
you are correct ..
you can use the gui for analysis and display of the histogram data per channel.
you will need to 'calculate' an appropriate 'sampling rate' to match the number of bins, and set the prestim baseline to zero
this should allow you to load the data in from a text file as if it were ERP data
we do not do any 'zeroing' of the ERP data with respect to onset/baseline.
nancy
Untitled Post
I'm Online
Posted on 06/26/09 11:27:36
Number of posts: 291
I don't think that you can visualize your command-line result with PLSgui directly, because "PLScmd program (see Appendix B) is a command-line PLS, and it is a separate program that is not compatible with PLSgui.", which is in the User's Guide.
However, I can help you to visualize your wave form to some extent, by twisting a dummy result file. Since there are too many manual work that is only for your data, it is not recommended for other people. If you could upload your command-line result .mat file to your ftp site, and send me the link, I am glad to help you.
Untitled Post
Posted on 06/26/09 11:35:05
Number of posts: 41
Nancy & Jimmy,
Thank you both for your replies.
I would like to try the GUI method first (b/c it's convenient for my data set up) and if this fails will move to command line.
And so have one more question about the baseline correction (the sampling rate i understand - and will match to the number of freq bins). Is it not the case that the software will attempt to 'detrend' the data to the baseline? If this is so - then I do need to include a dummy baseline data point - no?
It seemed from your reply that this is not something I need to worry about.
al.
Thank you both for your replies.
I would like to try the GUI method first (b/c it's convenient for my data set up) and if this fails will move to command line.
And so have one more question about the baseline correction (the sampling rate i understand - and will match to the number of freq bins). Is it not the case that the software will attempt to 'detrend' the data to the baseline? If this is so - then I do need to include a dummy baseline data point - no?
It seemed from your reply that this is not something I need to worry about.
al.
Untitled Post
Posted on 06/26/09 11:45:56
Number of posts: 41
actually no worries... your reply is quite clear :)
i was just surprised to read it given that the event related fMRI program will normalize to the event onset. but i suppose that makes sense given that the event related fMRI pulls out epochs from a time series, whereas the standard ERP program analyzed epochs that have already been extracted and averaged.
thanks!
i was just surprised to read it given that the event related fMRI program will normalize to the event onset. but i suppose that makes sense given that the event related fMRI pulls out epochs from a time series, whereas the standard ERP program analyzed epochs that have already been extracted and averaged.
thanks!
Login to reply to this topic.
