| I have two quick (hopefully quick anyway) questions: one regarding the
| coregistration of "low-res" EPI images to hi-res EPI images and the other
| regarding spatial normalization of SPGRs:
|
| For the co-regstration, i was wondering if it was possible to do a
| coregistration with a high resolution EPI (as the target image) without
| changing the resolution of the object (low-res EPI) image?
If you use a structural image as the target, and one of the EPI time series
as the object, then you need to also have all of the other images from the
EPI time series as "other". It works by estimating the rigid body
transformation that matches the "object" to the "target", and saving it
for the "object" and "other" images. The "target" is assumed to remain still.
If you "coregister only", then SPM will create or modify .mat files for all
the images that are moved ("object" and "other"). These .mat files store
the relative orientations and positions of the images, and can be combined
automatically with deformations estimated by spatial normalisation. Because
of this, you can spatially normalise all the images of a subject using
spatial norm. parameters derived from one of them, providing that they have
all been coregistered - but not necessarily resliced.
You should not normally need to reslice the EPI data to match the high
resolution image. The .mat files should be enough.
|
| For the normalization I have been successfully normalizing a series of SPGRs
| using the default settings. However, ne scan which is "normal" prior to
| normalization has serious aritifactis introduced after the normalization is
| applied. Specifically the front of the head is compressed. When only the
| affine transfomation is applied (no non-linear components) no apparent
| warping of the image is seen. With the application of the non-linear
| components (regardless of the amount of basis functions applied or the
| degree of regularization used) the artifact is re-introduced. Has anyone had
| seen this problem before? I am confused as to why all prior and subsequent
| scans normalize appropriately, but this one does not. Even more confusing,
| if i semi-automatically skull-strip the "problem scan" then run the default
| setting normalization, no warping is seen. Any ideas?
This is probably due to the orientation that SPM thinks the image is in.
Use the <Check Reg> button to see if it is registered with the other
images. If it is not properly in register, then the nonlinear deformations
will deform the wrong part of the image (i.e. apply deformations intended
for one region, to another region instead).
Best regards,
-John
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|