 |
 |
It should not matter what voxel sizes you use - providing that the estimated FWHM of the image (in the stats stage) is at least two or three times the voxel sizes. This is in order that the assumptions required by Gaussian Random Field Theory are OK. Higher resolution may be marginally better - but processing is slower, and extra disk space is required. -John | Could anybody possibly give me some advice regarding the spatial | normalisation of SPECT imaging data. The images have a voxel size of | 5.4 x 5.4 x 5.4, and I have presently been using the default values | given in SPM99. | | Are these the best variables to use? | | Which parameters have you found optimal for the spatial normalisation | of SPECT data? | | I would be grateful if anyone can help. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%