[MINC-users] mincreeample transformation problem

[Andrew Janke]

On 5 April 2013 02:08, Colin Shaun Hawco, M <colin.hawco at mail.mcgill.ca> wrote:
> That is almost the steps I am doing, only for the transform for EPI to T1, i am using mritoself, with a linear transform (which works on all other data sets except this one).
>
> Also, for resampling, things work if I resample using -like the original EPI file (which then should not change the file dimentions or start points). So this command gives a reasonable (but not useable on a practical level) output:
>
> mincresample -like epi_file.mnc -transform concat.xfm epi_file.mnc epi_normalized.mnc

Can you please explain what you mean by "things work" here? If you
load epi_normalized.mnc and your EPI_model in register they are
aligned or are you just meaning that the resulting file is still "all
there"?

> However, if I use -like EPI_model.mnc, the registration fails, and the data is shifted partially out of range. But in order to perform my group analysis, I need the EPI data in the space, with the dimensions and start points, of the EPI_model.

There should be no problem getting this to happen but I am now
starting to suspect that your EPI->T1 registration is off.

> I have the feeling there is a problem with translating/moving the origon, but I lack the sophistication in minctools to figure out how.

Origin location should not come into it. minctracc/mincANTS outputs a
transformation matrix that is unaffected by such things.


a