Hi David,
On Thu, Mar 10, 2011 at 2:44 PM, David Carmichael
<[log in to unmask]> wrote:
> Hi Vladimir,
>
> We use spm_eeg_tf, it has been updated in a more recent spm8 version. The
> input to the function is now different (you don't have .tf fields).
There are some backward compatibility mechanisms in place to be
removed in the future so code using the old function should still
work. Also the old version of the function is still available as
spm_eeg_tf5.
> It also
> seems seems to insist on the data being epoched - is this the case and if so
> is there a good reason for this?
>
Yes. The TF facilities in SPM are aimed at epoched data because that's
what we have a processing pipeline for. I suspect it'd be difficult to
use SPM for working with TF of continuous data because even if you
comment out some lines or trick SPM into thinking that the data is
epoched, the way the data is handled or visualized is not optimized
for this so you can easily run into memory problems or everything will
be very slow. However, if you are suggesting that the old code worked
for you, then you can keep doing what you were doing before with
spm_eeg_tf5.
> What would you recommend for continuous power analysis (make one long
> epoch?)?
One thing you can do is just split your data into lets say 2 sec
segments and treat it as epoched. If not then, I'm afraid you'll have
to write your own code, but perhaps you should look at Fieldtrip code
which is included in SPM8 as many building blocks you might need are
already there. There might be some Matlab tools available for this
kind of things but I don't know of any.
Best,
Vladimir
> Thanks
> David
>
> --
> David Carmichael
> Lecturer in Neuroimaging and Biophysics
>
> Imaging and Biophysics Unit
> UCL Institute of Child Health
> 30 Guilford Street
> London, UK
> WC1N 1EH
>
> Tel +44 (0) 2079052298
> Fax +44 (0) 2079052358
>
> http://www.ucl.ac.uk/neuroscience/Page.php?ID=12&ResearcherID=392
>
>
|