Dear SPMers
I am having problem with using segment for one particular subject. I am
trying to segment a T1-wighted image to give GM, WM and CSF. The outputs
always severely distorted (see attached images).
I cant work out what is wrong. I have about 200 T1w images all with the
same acquisition parameters, image dimensions, header information and
similar image intensities, which all segment perfectly well. It is just
this one subject which is problematic.
I have tried the standard segment and "New Segment" methods. I have
tried varying the bias FWHM and regularisation parameters. I have tried
using SPM5 and SPM8. I have tried adding a small amount of noise to the
images. I have tried applying a small smoothing kernel. I have checked
for NaNs and infs in the image and there are none. Nothing works. I am
completely lost as to where the problem is coming from or how to fix it.
Segmenting using FSL FAST works fine. Unfortunately I want to run DARTEL
later which requires a *_sn.mat or *_seg8.mat file to work with.
Basically to solve this problem, there are 2 possible approaches:
1. Find out where the distortion is coming from in SPM segment and fix
it (I have tired everything i can think of so far).
2. Find a way of artificially creating a *_sn.mat or *_seg8.mat file
that will work with the successfully segmented images obtained with FSL
FAST.
Any advice on doing either of these things would be very helpful.
Many thanks
Mark
--
Dr. Mark Drakesmith
Research Associate
Caridff University Brain Research Imaging Centre (CUBRIC)
School of Psychology
Cardiff University
Park Place
Cardiff
CF10 3AT
Tel: +44 (0) 29 2087 0354
Fax: +44 (0) 29 2087 0339
|