Dear Alex,
| i found out more about the problem of global values returned as 'NaN'. It
| turned out that 2 subjects' scans were 'empty' after normalization, that
| means all pixel values were set to zero, so obviously spm_global couldn't
| determine any meaningful global values for these scans. What is puzzling,
| however, is that this "infected" all other scans of the same study, so that
| all of them had 'NaN' for global value! I don't know if this could be
| related to the specifics of our SPM96 implementation or whether it is a
| true bug in the program. In any case, I thought you might want to know.
Thanks for letting us know.
The other scans will have been "infected" by the grand mean scaling,
which you must have selected to apply study by study (scaling of study
grand means). The study grand mean is the mean of all the global means
for that study, and will be NaN if any of the individual images returns
NaN. Each image is then scaled by a common factor such that the study
grand mean is that specified (usually 50 for PET). Clearly if the grand
mean is NaN, then so will the global means after scaling.
SPM displays the scaled globals. If you don't use grand mean scaling,
(which would be appropriate with quantitative data) then you'll see the
true pattern of NaN's.
Hope this helps,
-andrew
+- Dr Andrew Holmes [log in to unmask]
| -___ __ __ Wellcome Department of Cognitive Neurology - |
| ( _)( )( ) Functional Imaging Laboratory, Stats & |
| ) _) )( )(__ 12 Queen Square, Systems |
| (_) (__)(____) London. WC1N 3BG. England, UK |
+---------------------------------------http://www.fil.ion.ucl.ac.uk/-+
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|