Re: [Marsbar-users] [SPM] A question about % signal change in MarsBaR

[Matthew B. <mat...@gm...>]

Hi,

On Wed, Jan 20, 2010 at 3:04 AM, fenglu <fe...@fi...> wrote:
> Hi, Matthew. Thank you for help. I think I have got to know how to compute % signal change.
> First, when we compute fitted time course, the time course due to an event is averaged between sessions. Second, we got the value of top half signal for this time course as 'd'. Then 'd' is divided by mean of the session and we got a 's'. Last, we average the 's's across a session.
> All the calculations above is voxel-based.

Nearly - the fitted time courses are per-session, and then divided by
their own session means, then, if you've selected to get signal change
across sessions by combining events, these are averaged to get the
final result,

Best,

Matthew