 |
 |
> > Dear SPM Users, > > For a week now and thanks to your help I've been able to use SPMWin and SPM96nt for analyzing my data. I just have one concern that I wanted to expose to you : Normalization. > > My data consists of 18 axial slices of 64x64 points, so the resolution of my original data is 7.7 mm in z axis, 4.7x4.7 mm2 in the x-y plane. When I do the normalization after the realignment I got an image 68 in z axis, 79 in x axis and 95 in y axis with a voxel resolution 2x2x2. I used the template T2. > > My questions are > > 1. Is it correct to change so much the original data ? > 2. Is there other way to deal with the data, in other words, may I keep the original resolution and still be able to normalize the data. > > Please excuse me if these questions don't make sense and if you have any suggestion to orient my analysis, I would appreciate it indeed. > > Paula Munoz > Biomedical Engineering > rutgers University > Dear Paula, as Sergey has already replied, there is absolutely nothing to prevent you from using any voxel size, including the original one. However, I believe you may be mixing up voxel size and resolution. The resolution may be strongly coupled to the voxel size (and for fMRI typically is), but it may also be rather independent. By "changing" your original data to a voxel size of 2x2x2mm you will decrease your voxel size, but you will not increase your "resolution" (in fact you are likely to somewhat decrease it through the interpolation process). Since the inferential testing in SPM is dependent on resolution, which it calculates from the data itself, and not on the voxel size the "change" of your data does not introduce any problems. In fact one of the assumptions of the inferential testing is that the images are "smooth" compared to the voxel size, which will be fulfilled if you normalise your data to 2x2x2mm voxels. Bottom line; there are no problems with normalising your data to a 2x2x2mm voxel size. The subsequent analysis will be perfectly valid. If however you would for some reason like to use another voxel size SPM will do that for you. Good luck Jesper %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%