[Eeglablist] newtimef on continuous data

[Arnaud Delorme]

Dear Gethin,

newtimef should and has worked on computing and plotting continuous  
data (I just tried to be sure). The ITC will be equal to 1 for all  
time-frequency points (by definition) but this is not a problem. There  
is also an option in newtimef to define multiple baseline periods on  
continous data for computing significance between conditions.

Arno

On 14 avr. 09, at 19:10, Gethin wrote:

> Dear all,
>
> We are using newtimef to run a time-frequency decomposition of our  
> data. As we are interested in exploring low frequency (delta)  
> activity, the function automatically removes the first and last 1.7  
> seconds of data. As an alternative to changing the parameters of the  
> decomposition to include fewer cycles (or using very large  
> overlapping epochs), we wanted to run the analysis on continuous  
> data so that we only need to remove the first few cycles of each  
> block. I can’t seem to run newtimef on continuous data (perhaps  
> because it cannot calculate ITC?). One potential workaround would be  
> to pull out blocks of data as epochs (say 100second epochs) and run  
> newtimef on these “epochs”. This would still require some juggling  
> to get the true trigger values back in but would at least mean  
> running the decomposition on less overall data (although this seems  
> to reintroduce memory issues). Given that I can use the complex data  
> output to calculate the phase coherence myself can anyone suggest a  
> simple workaround to avoid apparently needlessly analysing epochs  
> that overlap with the previous 2 or 3 trials for every single trial?
>
> Many thanks in advance
>
> Gethin
> <ATT00001.txt>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20090602/92030afc/attachment.html>