First of all, excuse me for the last mail I sent to the SPM List, it was an
error
> | I have spatially normalized SPECT data encoded on 2 bytes on SPM SPECT
> | Template, and I have noticed that there are some negative values in the
> | resulting normalized data, especially in the borders of the image. Has
> | anyone already noticed that ? Is it possible or may I have made a
mistake
>
> Unless the values have been truncated at zero (not recommended as it
introduces
> bias), your original filtered back-projected images should contain some
negative
> values in the background. If these images are all positive, then spatial
> normalisation may introduce small negative values if you use sinc
interpolation.
> If you have very large negative values, check that the image format you
have
> is fully compatible with SPM.
>
Thanks a lot for such information, I have actually used sinc interpolation
and that's why there are some negative values (the min is relatively small :
around -20)
Nevertheless, I would like to make ROI measurements on different SPECT
normalized on SPM Spect Template. And to be able to compare such
measurements, I would like to normalize each measure by the total number of
counts in SPECT data : should I take into account such negative values to
assess this total number of counts ?? I think I should not ??
Best regards,
Christophe
*************************************************
Christophe GROVA
UPRES EA « Intégration de données multimédia en anatomie
et physiologie cérébrale pour l 'aide à la décision et l 'enseignement »
Laboratoire SIM, faculte de Medecine,
Université de Rennes 1
tel : 02 99 33 68 63
fax : 02 99 33 68 64
e-mail : [log in to unmask]
site : http://sim3.univ-rennes1.fr
*************************************************
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|