Hi Jonathan,
yes I am entering all sessions in the realign module but I am not
writing or re-slicing at that step. I am just producing the
realignment parameters that that I am entering as regressors into the
model.
I assume that's ok.
Michael
On Jun 25, 2009, at 12:13 PM, Jonathan Peelle wrote:
> Hi Michael
>
> When you run realignment, you should enter all scans for each subject,
> one session at a time. Afterwards they should all be registered with
> each other (both within and across sessions/runs).
>
> Hope this helps!
>
> Jonathan
>
> On Thu, Jun 25, 2009 at 6:08 PM, Michael Froelich<[log in to unmask]>
> wrote:
>> Volkmar,
>>
>> the only issue I am wondering about now is the registration of
>> different
>> sessions to each other.
>> I understand that the realignment produces the movement parameters
>> that then
>> can be used as regressors and that takes care of the movement
>> within a
>> session.
>>
>> But let's assume that the position of the participant differs
>> slightly from
>> session 1 to session 2.
>> Since the co-registration, in my example, was based on aligning the
>> anatomical image to the first image of session 1 only, how does hat
>> guarantee that sessions 2, 3 ...
>> are also coregistered (matched up) with the anatomical image
>> correctly.
>>
>> For example, if the first image of session 2 is rotated by 1 degree
>> compared
>> to the first image of session 1, wouldn't the normalization step
>> be somewhat off for session 2?
>>
>>
>> Michael
>>
>>
>>
>> Michael Froelich, MD MS
>> E-mail: [log in to unmask]
>> office: (205) 975-0145
>>
>>
>>
>> On Jun 25, 2009, at 6:35 AM, Volkmar Glauche wrote:
>>
>>> Dear Michael,
>>>
>>> the procedure you use looks perfectly ok to me. Realigning all
>>> sessions
>>> means that all functional images are in register with the first
>>> image of
>>> the first session. This image is also the reference in your
>>> coregistration step - thus registering the anatomical image to the
>>> functional time series. Therefore, you can apply the normalisation
>>> parameters from segment to these images (of course assuming that
>>> all of
>>> the steps above produced accurate results).
>>>
>>> Volkmar
>>>
>>> Dear SPMers,
>>>
>>> I am unsure about a batch script that I have been using and just
>>> want to
>>> make sure what
>>> I am doing is correct.
>>>
>>> I am preprocessing mutisession fMRI data of a block design.
>>> (1) First, I am obtaining realign parameters ("Realign: Estimate")
>>> for
>>> all sessions for a
>>> subject.
>>> (2) Then, I am coregistering the anatomical image (source image). As
>>> Reference image, I
>>> am entering my "Realign: Estimate: Realigned Images (Sess 1)" file
>>> via
>>> "Dependency".
>>> (3) Then, I am segmenting the subject's anatomical image.
>>> (4) Then, I am normalizing ("Normalize: Write") all sessions using
>>> all
>>> "Realign: Estimate:
>>> Realigned Images (Sess 1, 2, ...)" as "Images to Write" and the "DEP
>>> Segment: Norm
>>> Params Subj->MNI" option as "Parameter File".
>>>
>>> Therefore, for the coregistration step, I have used only Session and
>>> probably just one
>>> image out of that functional series to coregister with the subject's
>>> anatomical image.
>>>
>>> I am asking naively a few questions:
>>> As I am apparently only coregistering one image of one session to my
>>> anatomical image,
>>> should I be concerned about deformations? Should I not consider
>>> several
>>> sessions for the
>>> coregistration to my anatomical image? Am I correct in the
>>> interpretation that, at the
>>> normalization step, the parameters from the normalization of the
>>> segmentated
>>> anatomical image are used to normalize my functional (Session)
>>> images?
>>>
>>> These questions are basically to make sure I understand the process
>>> correctly.
>>>
>>> Thanks,
>>>
>>> <https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0906&L=SPM&P=213880
>>> >
>>
|