Dear Ivana,
Do you use robust averaging? Also, did you try to subsample your TF
data. With 10 sec trials I'm quite sure you don't need a wavelet every
3ms which is what you would have by default. In the new TF tool (which
will be available to external users in the next release) there is an
option to subsample the time axis when using wavelets. Perhaps this is
something you should try.
Best,
Vladimir
On Thu, Apr 22, 2010 at 4:08 PM, Ivana Konvalinka
<[log in to unmask]> wrote:
> Dear SPMers,
>
> I'm running time-frequency analysis on my MEG data - I have a total of
> 60, 10s-long trials/block and I'm looking at the 1:60 Hz frequency
> range. I then baseline correct using rescale, and finally average the
> data for each condition. There are 4 conditions, so ~15
> trials/condition in each block, and for some reason the averaging
> seems to take a really long time - probably about 6 hours, sometimes
> more, whereas the tf analysis takes about 20 mins. Any idea why that
> might be?
>
> I should also mention that I downsampled the data to 300 Hz.
>
> Thanks!
>
> Ivana
>
|