I tried using the script two different ways. In the first, I used it
exactly as you described, but where I was expecting values of around 2% BOLD
signal change, I was getting around .7, and when I checked the standard
deviation, it was incredibly high (higher even than the .7% signal change I
was getting!). In my second attempt, I used featquery to get its % BOLD
signal change, and then only used the last part of your script to adjust the
parameter estimates. This returned values as high as 145%!
The values I'm getting all seem to be in proportion to one another as long
as I use the same method, but I would like something that is not just an
On Tue, 17 Aug 2004 23:04:23 +0100, Joseph Devlin <[log in to unmask]> wrote:
>Did you use the effect script directly or try to interface it with
>Featquery in some fashion? The raw values from the script, by the way, are
>simply the adjusted parameter estimates (which scale according to the
>intensity values from your scanner) so these need to be divided by the mean
>signal intensity of your ROI over the time series. Depending on the
>version of the script you're using, this may be the last column of values
>in the output. If you multiply the result by 100, you should get % signal
>change (ranging from 0-100%).
>Another thing to look out for is that the script is written to use PE
>values. If you change it to use contrast images or zstats or anything like
>that, the values will be incorrect.
>Do any of these sound familiar? If not, we'll probably need some more
>details on this one.