John:
I have been segmenting MRI's using SPM-99 with similar problems, and have
not been able to find a solution,
Regards,
Rob McClure
-----Original Message-----
From: chris summerfield [mailto:[log in to unmask]]
Sent: Wednesday, July 04, 2001 5:57 AM
To: [log in to unmask]
Subject: segmentation query
john and other experts
we are segmenting a series of T1 images with SPM99. For some reason,
about half of the images do not segment properly, whether normalized or
not - mainly SPM does not seem to be able to distinguish well between
grey matter and CSF (ie the ventricles are typically assigned to the
grey matter). Looking at these images, it is evident that they have
lower mean intensity than the others, and changing the intensity and
contrast in the images in Analyze (level/width) solves the problem for
most. But there are a few which are still segemened badly. the
intensity values given SPM's display function for grey, white & CSF do
not differ greatly for those for which the segmentation works, either in
the absolute values or in the ratios between the 3. Is there a way to
play with the segmentation defaults, to impose a different intensity
threshold between grey and black, or something similar? or perhaps
another thing I could try?
thanks
chris summerfield
dept psiquiatria i psicobiologia clinica
universitat de barcelona
|