Hi Ella
Sorry for the confusion. To clarify:
You probably ran a GLM analysis before you started to do the PPI. In this analysis, a file called filtered_func_data was created - this is just your raw data post- BET, temporal filtering etc.
You will have extracted your ROI timeseries from this filtered_func_data. That is correct.
Now, when you come to run your PPI you can again start with your raw data and run all preprocessing. Or, you can enter the filtered_func_data as your input, and then switch off preprocessing. These should have exactly the same effect, except:
- Using the filtered_func_data will save you time
- However, if you do use filtered_func_dataa and skip preprocessing, you won't be able to use the motion parameters as regressors. This is a slight disadvantage but is not specific to PPI.
Jill
On Thu, Apr 15, 2010 at 8:41 PM, Ella Hinton <[log in to unmask]> wrote:
> Hi
>
> I am in the process of running a PPI analysis for the first time. I would be grateful for some clarification regarding what preprocessing to set up in the Feat Design from Jill O'Reilly's page.
>
> Under 3. Set up your Feat design
> It says:
> Go into Feat and load your data and set up your pre-processing as normal
>
> but then it says:
> If you are using the filtered_func from your GLM analysis as input, remember not to re-do any pre-processing steps such as BET, deleting volumes, or filtering.
>
> I have just extracted the timecourse from the seed ROI using the filtered func data and added it as an EV, as instructed. Does this mean I should turn off BET etc from the Feat model, even though it originally suggests to set up the preprocessing as normal? I've tried running it with one individuals dataset both ways and get different results, and I'm not sure which to go with for the rest of the data.
>
> Apologies if I've missed something, but would be very grateful for some pointers!
> With many thanks
> Ella Hinton
>
|