In the case of the MRI data:
I have defined the origin in all my scans to the AC. As a second step I notice that my images have different matrix sizes, i.e. voxel size of 1.2x0.9x0.9, matrix = 135x256x256; and the follow-up scans have voxels of 1.2x0.8x0.8, matrix = 120x188x188. For which I decided to reslice the image so to increase the matrix of the second one. I'm sending you a picture with an example before and after (attached image with the three MRIs).
When I started the longitudinal VBM, it works till the "Coarse affine Registration" when the error prompts, when I open the images they appear to be rotated and their origin has been changed, see attached image with the two MRIs. The actual origin has been changed to be in the lower left part of the matrix, but as I said the image oriantation has been swapped. See in image ime that origin is in the AC (coordenates in blue at the upper-left corner of each view) and that I've moved the cross to the AC in the VBM processed images. Could the problem be caused beacuse this rotations made by the VBM8 preprocess?
I'm not sure how to solve the problem, and If I still need to add some more field of view (increase the matrix size by adding zeros), how much would be optimal?
Best Regards,
Gabriel
> Date: Mon, 27 Jan 2014 21:14:10 +0000
> From: [log in to unmask]
> Subject: Re: [SPM] longitudinal VBM8 for non-MRI nada
> To: [log in to unmask]
>
> Dear Gabriel,
>
> On Mon, 27 Jan 2014 13:22:33 +0000, Gabriel Go.Es. <[log in to unmask]> wrote:
>
> >Hello all,
> >
> >
> >I'm looking for performing lognitudinal analyses on MRI and non-MRI data, but I'm not sure if it is possible to apply all the preprocessing as with VBM8 toolbox to the non-MRI data, this is PET adata.
> The segmentation in VBM8 is limited to T1-weighted data. Thus, you cannot apply it to PET data. I would suggest the following steps:
> 1. realign your PET data using the halfway registration implemented in VBM8
> 2. create average PET image
> 3. use average PET image to estimate spatial normalization parameters using PET template
> 4. apply normalization parameters to all realigned PET data
>
> >
> >Also I'm having some troubles with the VBM8 preprocessing on the MRI data, I'm always getting the following error:
> >"There is not enough overlap in the images to obtain a solution.
> This message means that you probably have a limited field of view and the brain is not fully covered. In that case it is difficult to estimate normalization parameters. You can try to define the origin and roughly correct for rotation according to the template. Then, you might have better starting values. However, if your data are consisting of only a few slices this will be ot working, because information is insufficient to estimate any transformations.
>
> Best,
>
> Christian
>
> >
> >In both cases, MRI and PET I have cross-sectional data already VBM8 analyzed. I don't know if the preprocessing of this be valuable for any purpose?
> >
> >
> >Many thanks in advanced,
> >Gabriel
> >
> From: [log in to unmask]
> Subject: Re: [SPM] longitudinal VBM8 for non-MRI nada
> To: [log in to unmask]
>
> Dear Gabriel,
>
> On Mon, 27 Jan 2014 13:22:33 +0000, Gabriel Go.Es. <[log in to unmask]> wrote:
>
> >Hello all,
> >
> >
> >I'm looking for performing lognitudinal analyses on MRI and non-MRI data, but I'm not sure if it is possible to apply all the preprocessing as with VBM8 toolbox to the non-MRI data, this is PET adata.
> The segmentation in VBM8 is limited to T1-weighted data. Thus, you cannot apply it to PET data. I would suggest the following steps:
> 1. realign your PET data using the halfway registration implemented in VBM8
> 2. create average PET image
> 3. use average PET image to estimate spatial normalization parameters using PET template
> 4. apply normalization parameters to all realigned PET data
>
> >
> >Also I'm having some troubles with the VBM8 preprocessing on the MRI data, I'm always getting the following error:
> >"There is not enough overlap in the images to obtain a solution.
> This message means that you probably have a limited field of view and the brain is not fully covered. In that case it is difficult to estimate normalization parameters. You can try to define the origin and roughly correct for rotation according to the template. Then, you might have better starting values. However, if your data are consisting of only a few slices this will be ot working, because information is insufficient to estimate any transformations.
>
> Best,
>
> Christian
>
> >
> >In both cases, MRI and PET I have cross-sectional data already VBM8 analyzed. I don't know if the preprocessing of this be valuable for any purpose?
> >
> >
> >Many thanks in advanced,
> >Gabriel
> >