Hi all,
I've got a slightly odd result from SPM5's bias correction, the
attached screenshot shows six images followed by their bias corrected
(m...) versions at the bottom. These images appear to have fairly
uniform intensity to start with, and the chief effect of the bias
correction has been to darken them.
However, the fourth image has not been darkened to the same extent as
the rest, with the result that its m image (bottom left of fig) has
now become an outlier with respect to the others. Can anyone help to
explain this result? Thanks.
I am intending to do some further investigation of the images using
the voxel intensities themselves (as well as parallel work on the SPM5
segmentations), and I hoped that the bias corrected images would have
more meaningful intensities. It appears from this that I might be
better off sticking to the originals, as the non-uniformity seems
minor, while the generation of apparently random outliers seems worrying.
I'd appreciate anyone's thoughts on this, and perhaps any
recommendations of other bias correction software. Thanks,
Ged.
|