---------- Forwarded Message ----------
Subject: Re: segmentation after coregistration
Date: Friday 12 March 2004 12:56
From: John Ashburner <>
To: Torsten Ruest <[log in to unmask]>, [log in to unmask]
Higher resolution images can be segmented better than low resolution images.
I would suggest segmenting the images before you reslice with AIR (ie reslice
the grey and white matter images).
Best regards,
-John
> I am trying to segment MRI images with SPM after they have been
> coregistered to PET images (using AIR). The dimensions of the MRI (skull
> stripped off) before coregistration are 144*256*151, and those of the PET
> are 128*128*47. The coregistration seems to work fine. For segmenation via
> SPM, the transformation matrix obtained from AIR has been applied to the
> original MRI (not stripped off skull). Unfortunately, the SPM segmentation
> step gives not the desired results.
>
> Then I tried to coregister the original MRI (not stripped off skull) with
> the PET images using SPM, but this attempt gives false results (brain size
> to large, not the whole brain in the image etc).
>
> I am wondering, if the amount of slices (47) in the coregistered image and
> in the PET image are enough to give good segmentation and coregistration
> results, respectively.
>
> Any help about how to solve the segmentation problem is appreciated.
> Many thanks in advance.
-------------------------------------------------------
|