Hi there,

we would like to use cmtk's epiunwarp facility on fmri data that have been acquired by the Human Connectome Project (HCP).

The data consists of 2 runs, acquired with opposite phase encoding. My understanding of epiunwarp is that only 2 images with opposing phase encoding direction are needed, with the remainder of the data being  acquired using one of the 2 schemes.

So my question is whether I can just use the data of the 2 runs as a whole, eg pool the ~200 volumes in one direction with the other 200 acquired in the other direction, or if they need to be treated separately (in the latter case additional, separate images could be used which are provided by HCP, serving as opposing phase encoding images).

Thanks very much in advance.

Best wishes, Torsten

sorry, just ignore my previous message, after giving it some thought I am sure that they need to be treated separately. Cheers,

Okay, will ignore...

Except, you may want to read the original paper by Holland on the reverse phase-encode business. They are saying there that the scheme cannot be applied to BOLD data because of non-warping-related intensity effects (or so I vaguely recall).

Best,
  Torsten

To make another comment on this one:

I just noticed that epiunwarp outputs only the forward deformation field. In your application, you probably want the reverse field also, because otherwise you'd have to run epiunwarp twice, which would be ridiculous (and potentially inconsistent).

I should probably add optional output of the reverse field.

What do you think?

Okay, as of SVN r5333, epiunwarp can optionally also export the reverse deformation field.

(Note that until I put them back in, the SVN trunk is missing the correct_dwi_XXX scripts.)

TR