Thanks, no worries. I was just being curious. Kind regards, Michel Sent from my iPad On Jul 25, 2012, at 12:23 PM, Mark Jenkinson <[log in to unmask]> wrote: > Dear Michel, > > Creating a halfway transformation for fnirt is a *lot* harder and we have not implemented this. > So I'm afraid the answer at the moment is no. > > All the best, > Mark > > > On 25 Jul 2012, at 12:21, Michel Thiebaut de Schotten wrote: > >> Dear all, >> >> Would it be possible to carry on the same analysis with fnirt? >> Thanks for your response, >> >> Michel >> >> Sent from my iPad >> >> On Jul 25, 2012, at 10:42 AM, Mark Jenkinson <[log in to unmask]> wrote: >> >>> Hi, >>> >>> Yes, in SIENA we just use the one transformation. >>> Essentially we do the same thing as in the script - use avscale to get the halfway transforms. >>> It outputs both a forward halfway and a backward halfway transform, so you can get all you need from one call of avscale on a single transformation matrix. >>> >>> All the best, >>> Mark >>> >>> >>> On 25 Jul 2012, at 10:37, Andreas Bartsch wrote: >>> >>>> Hi, >>>> >>>> that would assume that the registration is inversely consistent between >>>> the two images which is not necessarily the case. >>>> Cheers, >>>> Andreas >>>> >>>> Am 25.07.12 11:20 schrieb "Gabor Perlaki" unter >>>> <[log in to unmask]>: >>>> >>>>> Dear Saad, >>>>> >>>>> Thank you for your answer. I don't understand something in the script >>>>> you sent. >>>>> >>>>> There is two registration in the script: >>>>> >>>>> 1, A image to the B >>>>> 2, B image to the A >>>>> >>>>> and after that the Forward half transform from the first registration is >>>>> applied for A image and the Forward half transform from the second >>>>> registration is applied for B image. >>>>> >>>>> Why do we need 2 registrations? Wouldn't it better to do only the first >>>>> registration and after that applying the Forward half transform to the A >>>>> image and the Backward half transform to the B image? >>>>> >>>>> Thanks, >>>>> Gabor >>>> >>