 |
 |
I checked already the orientation of the image, of the lesion and of the template and it looks perfect. The format of the image is *.nii. Is that what you mean? best, Giulio ________________________________ Da: John Ashburner <[log in to unmask]> A: Giulio Pergola <[log in to unmask]> Cc: [log in to unmask] Inviato: Mar 2 marzo 2010, 15:54:25 Oggetto: Re: [SPM] Normalization using masks Ensure the file format of your image is correct, and that they are roughly aligned with the template before you begin. You can use Check Reg to do this, but you may want to manually reorient your images using the options in Display. Best regards, -John On Tue, 2010-03-02 at 14:16 +0000, Giulio Pergola wrote: > Dear Prof. Ashburner, > > thank you for your help, what you suggested worked. I used the lesion > image as a mask for the normalisation and the T1 template as a > template. Unfortunately, the result of the normalization is not > satisfactory: the brain is somehow rotated by 45° and completely > deformed. What might have happened? > I thought that, since the *.nii of the lesion is white on black > background, I have to revert the values for the VOI; but in that case > SPM8 complains that there is not enough overlap between tha images. > How can I tackle this problem? > > best regards, > > Giulio Pergola > > Inst. of Cognitive Neuroscience, Department of Neuropsychology > Ruhr-University Bochum GAFO 05/611 > Universitätsstraße 150, 44801 Bochum, Germany > Phone: +49-234-3223175 > Fax: +49-234-3214622 > > > > > ______________________________________________________________________ > Da: John Ashburner <[log in to unmask]> > A: Giulio Pergola <[log in to unmask]> > Cc: [log in to unmask] > Inviato: Lun 22 febbraio 2010, 12:39:04 > Oggetto: Re: [SPM] Normalization using masks > > > > > > I have several T1 scans of patients with lesions close to the > > ventricles, and I am trying to normalise them. > > I would like to follow both 1) the cost function masking and 2) the > > segmentation approaches and compare the results. > > > > 1) I encounter problems when I try to Normalise (Est&Write) using > > a .nii mask drawn with MRIcron. Here is the error message: > > > > Fine Affine Registration.. > > Failed 'Normalise: Estimate & Write' > > Error using ==> spm_affreg at 67 > > Source and its weighting image must have same orientation > > Try reslicing the weighting image to be in voxel-for-voxel alignment > with the image that is warped into alignment with the template. You > can > do this with coregister, reslice only. I'm guessing that the mask > images are flipped relative to the MR image. > > > > > > > 2) I am getting a bit confused about what is the best procedure for > an > > automated normalisation. In Crinion et al., 2007 I read it is best > to > > segment the images; in the SPM mailing list Prof. Ashburner > > recommended to do it through the New Segmentation method; in Seghier > > et al., 2008 it seems you can somehow have the program automatically > > draw the lesions. > > The Seghier et al work is not included as part of SPM. I would expect > that the new segmentation model would do a better job than the older > version, but I may be wrong. As ever, it is an example of selecting > the > best model of the data, which generally needs to be done empirically. > > Best regards, > -John > > -- > John Ashburner <[log in to unmask]> > > > -- John Ashburner <[log in to unmask]>