Dear Aidan,
On Thu, Nov 25, 2010 at 2:49 PM, [log in to unmask] <[log in to unmask]> wrote:
> 1. When using spm_eeg_tf, what units of measurment are used in the power
> output and is there any obvious way to get from this to raw amplitude (i.e.,
> fT)?
The units are the original units squared which would usually be uV^2
for EEG and T^2 for MEG (quite small numbers). When exporting to
images we scale the MEG data to fT^2 to avoid numeric issues. In the
near future we'll probably switch to working in fT for MEG throughout
SPM.
> 2. Is there any particular means of rescaling or baseline correcting that
> you would advise? I am using the LogR option at present (epoched from -700,
> but using a -200:0 baseline), but only because it makes the results easier
> to eye-ball across different frequency ranges. Is these any sense of taking
> the log of each frequency band, and then baseline correcting with a simple
> subtraction?
This will be exactly what LogR does as difference of the logs is log
of the ratio. In general it's a matter of taste and what is most
sensitive to your effect of interest. LogR and Rel are quite similar,
but I was once asked by a reviewer to use the latter, but it's
probably related to what is more common in a particular community.
What you see when you baseline-correct might be quite
different/complementary to what you see when you just take a log so
you might want to look at both to get the full picture.
Vladimir
>
> Any help/advice appreciated,
> Aidan
>
> --
> ====================================
> Dr Aidan J Horner
> Institute of Cognitive Neuroscience
> University College London
> 17 Queen Square
> London WC1N 3AR
>
> Tel: 020 7679 4727
> Fax: 020 7813 2835
>
> http://www.icn.ucl.ac.uk/ahorner/
> ====================================
>
>
|