> It is my understanding that in SPM5 bias correction is applied to all
> the images before they get segmented, whether you choose to save the
> bias-corrected image or not. Is this correct?
Yes, as the bias correction significantly improves tissue classification
by modelling smooth intensity variations in the image.
> Also I found that with some of my images segmentation results improve if
> I take the saved bias corrected images and segment these instead of the
> raw images. If I understand this correctly, these images get
> bias-corrected twice. Is this OK to do?
Your understanding is correct. I am not sure this is the right thing to
do but ages ago, using spm99, I did the same thing with inhomogenous
data from a 3T. My reason for doing this was that the initial affine
registration did not work in some cases in the original images but did
in the corrected ones. If your two-step-approach significantly improves
your segmentation results it may be worthwhile to give the bias
correction more freedom in the first place, i.e., play around with the
settings of bias cutoff and regularization in the segmentation options.
But if that does not work I don't think it is a sin to do it your way ;)
although there may be theoretical disadvantages to this that I do not see.
Marko Wilke (Dr.med./M.D.)
[log in to unmask]
Universitäts-Kinderklinik University Children's Hospital
Abt. III (Neuropädiatrie) Dept. III (Pediatric neurology)
Hoppe-Seyler-Str. 1, D - 72076 Tübingen
Tel.: (+49) 07071 29-83416 Fax: (+49) 07071 29-5473