Dear Ying, If there was a "better" (how ever you define this betterness) measure of the match between a pair of images, this measure should actually be employed to match the images instead of using a "less good one". :-) Visual check just ensures that the realignment/coregistration did not completely go wrong. You can automatize this by generating CheckReg images and saving them in different pictures or create a movie of them all. This saves the hassle of manually clicking for each image. For the realignment what matters really is the volume-to-volume displacement, i.e. speed, so you could derive a (few) scalar(s) to summarize this over the whole time series. I am sure there are some tools out there already calculating this kind of measure. HTH, Chris Le 12/01/2019 à 19:51, Liuying L. a écrit : > Hi SPM experts, > > I found the method of checking registration manually with the naked > eye (by pressing the 'Check Reg' button on SPM GUI) is very inaccurate > and difficult. > > Do you know SPM functions or the other functions which can > automatically evaluate the quality of the alignment between functional > images, or the registration between anatomy and functional images? > Or do you know more accurate and easier ways to check registration? > > Best Regards, > Ying