Hello Chris,
finally the problem was due to the SPM 5 updates from 18th january. That
was the answer of John Ashburner:
"How were the scalefactors in the original lesion map images set, and what
value was assigned to the lesion in these data? If a value of 1 was
assigned
to the lesion (before accounting for the scalefactor), then the resliced
data
would require values between 0 and 1 to represent the intensities. These
would be rounded to the closest value (ie zero or one).
In previous versions (before these updates), SPM automatically rescaled the
intensities to the max and min values in the volume and determined a
suitable
scalefactor. This strategy caused problems for 4D data (see emails on
list),
so was changed. The current strategy is now to preserve the original
scalefactor.
If you have an 8 bit lesion mask, then I would suggest setting the lesion
voxels to a value of 255, and using a scalefactor of 1/255."
The first time I multiplied the normalized images with 255 but it did
not work.
But in fact when multiplying the images and masks of the lesions before
normalization, it works.
Thanks a lot
Elisabeth
|