Dear List,
On Monday I posted a question concerning the application of previous
transforms on 4D images (see appended). After searching the archives, I
found John's excellent post concerning changing the spm_default.m "multivol =
true" which saves me ONE multivolume parse/operation/recombine step.
Using 3rd party software I was able to construct a coregistered multivolume
file. THEN, after changing the spm_default.m, I was able to spatially normalize
my coregistered "sum" image (works fine) and when I selected "Write
Normalised Only" on the 3rd-party 4D file, a tidy list of individual volumes came
up in the GUI and I selected them in order, hit "Done" and SPM wrote out a
perfectly servicable 4D, spatially normalised file.
Thinking "excellent... lets see if I can cut out the 3rd-party step", I started
from scratch, and I coregistered "sum" image in SPM afresh. During the initial
interaction with the GUI, I selected "Other Images" > "the uncoregistered 4D
file", hit "Done". Unfortunately, as happened on Monday, SPM only wrote the
first volume (1 of 21, pretty much useless on it own).
OK, no good. Lets try applying the transform after the fact...
Started from scratch again, but this time I only coregistered and wrote out
the "sum" image. With the *.mat file in place, I restarted the SPM coregister
routine again and selected "Reslice Only" > "Space defining image"
> "coregistered-sum-PET", thinking the change in spm_default.m would engage
as it had in the spatnorm test I'd run earlier and I'd see a similar list of
individual volumes to select. No such luck. Also, only the 1st volume wrote
out again.
While I actually have a solution I could live with (albeit still a little clunky), I'd
really like to avoid the 3rd party solution:
---Is there some reason why "multivol = true" works with spatial writing step
and not the coreg writing step?
---Is there something in the spm_coreg.m I could/should change?
Thanks in advance,
Randy
PS. Alexander: Thank you for your speedy reply on Monday.
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
John's 2005 post
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
Subject: Re: dynamic spm2 analyze format
From: John Ashburner <[log in to unmask]>
Reply-To: John Ashburner <[log in to unmask]>
Date: Wed, 12 Oct 2005 10:42:47 +0100
Content-Type: text/plain
> I have a question about an Analyze format and SPM's capability to handle
> it.
>
> Does Analyze allow us to store 'dynamic' images (images at different time
> points) in a single file? Can SPM2 handle this kind of file, especially in
> the co-registration or spatial normalization?
>
> We are having a need to register and warp dynamic PET images using SPM2.
In principle, SPM2 can handle 4D volumes, although you need to set the
defaults.analyze.multivol field to enable this. This is either by editing
spm_defaults.m, or by starting up SPM and typing..
global defaults
defaults.analyze.multivol = true;
SPM2 has a few slight problems in terms of file selection though, but these
should be resolved in spm5b. In SPM5b, when selecting images, you would
change the field that normally says "1", to the range of volume numbers you
want (e.g. "1:30").
Reading the image headers in order to obtain the number of volumes has a bit
of an impact on speed, so you don't by default have the option to select
individual volumes.
Best regards,
-John
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
My Monday post and Alexander's reply
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
RE: [SPM] Coregistration (reslice only?) of multiframe image after summed-
frame image works OK
Monday, December 8, 2008 9:40 PM
From: "Hammers, Alexander" <[log in to unmask]>Add sender
to Contacts To: "Randolph Andrews" <[log in to unmask]>,
[log in to unmask]
Dear Randolph,
I'm afraid when we did this we did it to individual frames, too. Maybe
you can script this to be ready before this year's Christmas?
Have you thought of another much more pain-free option - work in PET
space and coregister the MRI to your summed PET image?
Good luck & hope this helps,
Alexander
PS: Incidentally, you should normally weigh your frames for your summed
image depending on frame length and number of events - but for a simple
task like co-registration your simple sum will work for most tracers
with any cortical signal at all.
-----Original Message-----
From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]]
On Behalf Of Randolph Andrews
Sent: 08 December 2008 21:03
To: [log in to unmask]
Subject: [SPM] Coregistration (reslice only?) of multiframe image after
summed-frame image works OK
I'm very used to working with static images (FDG using SPM96, SPM99, SPM2)
but I now need to coregister dynamic images (21 frames) to a subject's MRI.
The early frames don't have a whole lot of information, so I've summed all
frames into a sngle image, and the coregistration works just fine on that.
I originally thought I could coregister and reslice the summed image and when
the GUI prompted for "other images", I'd just select the dynamic image and
SPM would use the transform obtained from the first coregistation and reslice
all 21 frames of 2nd image.
It appears to only reslice the 1st frame of the dynamic image (not a whole lot
of use by itself). I've seen some earlier posts (4D image) that suggested I
parse the dynamic image frame-by-frame, but I have 96 of these, and I'd like
to get this coregistration step done before Christmas (of next year).
Using "Reslice Only" prompts me to select "Space defining image, subj 1". I've
tried both the r(summed image) and the MRI, and I still only get the 1st frame.
Help?
|
