 |
 |
Dear John & SPM > I'm afraid I don't fully understand what you are doing. The origins > in normalised (nr*.img) images should differ from those of the > (r*.img) un-normalised images. there is obviously something fundamental I have misunderstood. When you say that normalisation changes the origin location, do you mean that it changes the voxel coordinates but the anatomical location remains the same, or vice versa? I change the origin on the r* files so that it is visually (using check reg) in the same spot as the origin of the EPI template. Following normalisation, the origin of the nr* files is no longer in register with that of the template - does this mean that normalisation has not occurred correctly? Should I change the origin again to be in the same place as the EPI template before smoothing and estimating the model such that the activations in the glass brain are appropriately located? Chris Summerfield departament de psiquiatria i psicobiologia clinica universitat de barcelona > Best regards, > -John > > | we have a question about the output of the normalisation process for > | fMRI. We have noticed that the origin of the normalised nr* files differs from > | that of files at earlier stages (eg r* files). Despite the fact that > | we carefully change the origin of the files to equate with that of > | the EPI template image (using header edit) following normalisation > | it reverts to a more superior location and our normalised images turn > | out "squint". > | > | We suspect that this has to do with the fact that during > | normalisation, SPM reads origin information from the .mat files and > | not the header (is this right?). We cannot delete the .mat files, because these > | contain information about rotations which we have performed on the > | images such that the slices are at the correct angle. Our question > | is this: can you persuade SPM to read the origin location from the > | header without deleting the .mat files? > | > | the only solution we have come up with so far is to physically move > | the images (using the up, left, etc tools in the display window) until > | they align with the origin, rather than vice versa. Is this the only > | way to do it, or is there something we have misunderstood? >