I should clarify what I just said. The normalise button does not work
as well as the segment button for estimating the spatial normalisation
parameters, but if you actually want to write spatially normalised
images, then you need to use the Normalise Write option - after
estimating the parameters with the Segment button.
Best regards,
-John
On 4 October 2010 19:03, John Ashburner <[log in to unmask]> wrote:
> For MRI, I would not really recommend using the Normalise button to
> spatially normalise. If you use the segment button, then this will
> write out some parameters that can be used for spatially normalising
> your data. In principle, the "New Segment" toolbox should do a better
> job than the algorithm you get from the segment button, but I have to
> do a bit more work in terms of integrating the deformation fields that
> this produces into SPM.
>
> There's a paper by Arno Klein et al in NeuroImage, which shows how
> poorly the Normalise button does compared to other approaches
> (including other approaches available in SPM). A paper by Crinion et
> al also shows that the Normalise button does not work very well when
> compared with the Segment button.
>
> Best regards,
> -John
>
> On 4 October 2010 18:46, Asaf Kaftory <[log in to unmask]> wrote:
>> Hello dear SPM'ers,
>> I have two questions regarding my pre-processing batch file:
>>
>> 1. Under Normalization, should the "source image" be "Bias corr" (a dependent product of the segmentation phase) ? - I thought i should supply a *.mat parameter file that is the product of the segmentation process, however we don't have an insertion of parameter file option and besides, that *.mat file hasn't been created yet (since this is the batch configuration which we have yet to run + it doesn't appear in the dependencies) ?
>>
>> 2. Under the anatomic Normalization, what should the "images to write" be ? the same as the source image ("Bias corr") ?
>>
>> thanks,
>> asaf.
>>
>
|