I am attempting to normalize skull-stripped structural MRI images to each other in such a way that I will be able to use the .mat deformations to get at some deformation based morphometry. I am having quite a bit of trouble getting the images to normalize correctly, and my brains are getting cut-off, and in some situations are completely outside of the bounding box, when I attempt to use increase the number of basis functions. One suggestion that I had to get around this problem was to normalize to a few basis functions before preforming a second normalization in in an attempt tohave the starting points for my later estimates, across other parameters, closer and have less chance for error. The problem with this approach is that it requires a reslicing upon each normalization, which as I understand, will add some smoothing to my images. I know that the .mat file contains the information such that the reslice is not nessisary, but I am wondering if I will be able to use the .mat file as a starting point for a completely new normalization... Is there a way that I can tell SPM to run 4x4x4 basis functions, and keep the result as the groundwork for further normalization attempts? The documentation seems to be claiming that there is a way to do this, but I have been unable to get such a thing to function. Beyond that, if anyone knows why my normalisation attempts are working out so poorly and/or if there is a way to improve them, I would be in eternal gratitude. Thank you very much for your help, Andrew Fox