Hi Bas,
Thx for the info. I was also thinking about the two options you
provided, but I think I will stick with the first one because the
movement is over a large period (see attached plot, TR=3s).
Are there other options in this case?
Pieter
Op 21/09/2012 13:34, Bas Neggers schreef:
> Hi Pieter,
>
> you could add a dummy regressor for each block where subjects moved
> (0, and 1 while the bad block was on), and not model the block in your
> regressors of interest.
>
> Alternatively you could replace the bad scans with an interpolated
> estimate of neighbouring 'good' scans, I think Artrepair does
> something like that.
>
> Goog luck,
>
> Bas
>
> On 09/21/2012 01:16 PM, Pieter Vandemaele wrote:
>> Dear SPM'ers,
>>
>> I have data from a two-condition block design.
>> The subject moved too much in two blocks, one in each condition.
>> What is the best way to remove them from analysis?
>>
>> Best
>>
>> Pieter
>
>
|