Dear Otto,
For the two approaches to return the same time series, you need to
specify the same adjustments. To do so, specify an F-contrast over the
effects of interest (probably everything but motion regressors and
constant terms) then, either:
1. Plot: select the F-contrast and choose "Adjusted response"
2. Eigenvariate: select the F-contrast when asked "adjust data for..."
Variables y from Plot and xY.u from Eigenvariate should then match.
Best regards,
Guillaume.
On 03/09/17 04:47, Otto Muzik wrote:
> I have a question regarding extraction of fMRI time series':
>
> As I understand, there are 2 ways to extract a time series once
> I have selected a significant cluster using the Results tab:
>
> 1. use the plot feature, then use "Fitted Response", select the contrast,
> select "Predicted response" and then plot "against time".
>
> 2. specify co-ordinates and select the "eigenvariate" button, select
> "don't adjust" and pick a 2x2x2 box (1 voxel).
>
> Here are my questions for which I could not find any answers:
>
> I assume that both these time series' are corrected for patient motion
> and scanner drift, is that correct?
>
> Why are these time series' different, given that I am using both times
> the same voxel? There can be only one time series once the data is
> corrected for nuisance variables (patient motion and scanner drift).
>
> #1 will be a function between -2 and 2 (I assume in % of signal change).
> Is that correct?
>
> #2 will be a function between 130 and 180. Why?
> For what is #2 "adjusted"?
> #2 will also have outliers and the first and last time point which need to
> be corrected before correlating time series'. Why are outliers present?
>
> I would really appreciate if somebody could point me in the right direction.
> I've been thinking about this for a while but I am really stuck.
>
> Many thanks!!
>
> Otto
>
--
Guillaume Flandin, PhD
Wellcome Trust Centre for Neuroimaging
University College London
12 Queen Square
London WC1N 3BG
|