Dear Christine,
On Thu, 22 Sep 2011 01:54:31 +0100, Christine G <[log in to unmask]> wrote:
>Dear John,
>
>I'd like to integrate the VBM output with Dartel Normalize to MNI for normalizing the functional images, rather then apply deformation fields (the combined smoothing seems to be a good idea). I know I could do Dartel Create Template to generate the flow fields. but it takes a long time. Is there a way to transform the deformation field (VBM8 output) to flow fields that is the required input for Normalize to MNI?
I am not aware of any way to easily convert the deformation fields to flow fields. However, you can export the VBM8 segmentations to Dartel (I suggest to export the affine transformed segmentations) and use Dartel with an existing template. The template provided with VBM8 is based on 550 healthy control subjects of the IXI database and is already transformed to MNI152 space. However, I am not sure whether the use of an existing template is soo much faster...
Regards,
Christian
____________________________________________________________________________
Christian Gaser, Ph.D.
Departments of Psychiatry and Neurology
Friedrich-Schiller-University of Jena
Jahnstrasse 3, D-07743 Jena, Germany
Tel: ++49-3641-934752 Fax: ++49-3641-934755
e-mail: [log in to unmask]
http://dbm.neuro.uni-jena.de
>
>Thanks
>Christine
>
>
>>>>>>>>>>
>Image registration with Dartel uses many more degrees of freedom than
>the old basis function approaches. For example, using 8x10x8 basis
>functions would correspond to fitting a model with only 1,920
>parameters, whereas Dartel typically has three times as many
>parameters as there are voxels in the "imported" data (using about
>3,000 times as many parameters). This means that it can fit parts of
>the brain that lower-dimensional approaches can not. For this reason,
>I would expect the warped images to be a bit rougher as there is much
>more flexibility to expand and contract different regions.
>
>To check that the procedure you are using is good, I would suggest:
>1) Use Check Reg between the original fMRI and anatomical scans to see
>if they are well aligned.
>2) Apply the same procedure to the anatomical data.
>
>Whether or not the extra anatomical alignment accuracy will translate
>to extra accuracy for aligning functional information will largely
>depend on how well the native-space fMRI and anatomical scans were
>registered. If this alignment is not good (because of the usual
>distortion issues), then estimating the more detailed deformations
>from the anatomical scans will not help.
>
>In fact, with a simple resampling scheme, the extra detail may even
>lose you some sensitivity, because it can mean that signal from some
>of the original voxels can be lost. For this reason, the normalise to
>MNI space option of Dartel tries to be a bit more intelligent. By
>combining the warping and smoothing in such a way that the values in
>the output data represent the actual average signal intensity from the
>original scans, the amount of signal lost should be reduced.
>
>Note that this approach does cause those streaky effects in fMRI data
>without full brain coverage. The streaky effects arise because the
>warped images contain the weighted average signal intensity over
>regions defined by the smoothing kernel. This means that for regions
>of the spatially normalised data that lie outside the FOV of the
>original fMRI, the average intensity is largely computed from the
>closest available region of the original image.
>
>Best regards,
>-John
|