Some thoughts:
1) DARTEL calculates mutual warps to a midpoint. If you enter Time1 and Time2, it produces flow fields from each timepoint to a calculated template halfway between Time1 and Time2. If you apply the DARTEL flow fields only to Time1, the image will not be registered with Time2. Btw I think mutual warping is great - it means there is no bias of one timepoint always driving normalization/segmentation.
2) That said, your picture looks worse than it should even if it was just warped to the subject-specific template. What did you apply u* to? The flow fields (u*) apply to images in the original native position. Create Warped is smart enough to work when you apply u* to the DARTEL-imported images, but if you try to apply u* to anything other than the DARTEL-imported segments or images the original position, the warp will not match what it's being applied to and will end up looking odd.
-Dana
-----Original Message-----
From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On Behalf Of Jamie Hanson
Sent: Saturday, June 06, 2009 2:57 PM
To: [log in to unmask]
Subject: [SPM] DARTEL warping issue?
Hi SPM folks-
I was trying to do some longitudinal structural imaging processing w/
DARTEL, but kept running into some issues. I had two waves of
structural data (2 High-res T1s) and I wanted to warp a subject's Time
1 scan to their Time 2 scan and then create a Time 2 template (which
all the Time 2 scans would be warped to). However, when I warped a
subject's Time 1 scan to their Time 2 scan, the output get shifted
down (and are greatly misregistered).
Examples:
http://tezpur.keck.waisman.wisc.edu/~hanson/GM_original.png
-Time 1 data
http://tezpur.keck.waisman.wisc.edu/~hanson/GM_warped_dartel.png
-Time 1 data warped to Time 2
Neither the raw Time 1, nor the raw Time 2 data is that far down on
the screen. The registration algorithms seem to pushing things down.
Any thoughts on why this would be happening?
Best,
jamie.
--
Jamie L. Hanson
Waisman Laboratory for Brain Imaging & Behavior | Child Emotion Research Lab
University of Wisconsin - Madison
1500 Highland Avenue
Madison, WI 53705
Email: [log in to unmask]
Homepage: http://tezpur.keck.waisman.wisc.edu/~hanson/
|