Dear All,
apologies for sending this again but I am still hoping to find an
explanation for the observation described below. I tried other series
from other patients and processed the same series in spm5 entirely, with
similar results.
If nobody else sees it, I will happily take all hints how to correct my
mistake, but if others see this too, I wonder what the opinion is as to
where this difference stems from. As always happy to be corrected, and
thanks in davance,
Marko
-------- Original Message --------
Dear All,
now that I finally got started :) I used spm5 to re-analyze a set of
functionals that was processed and analyzed before in spm2 (I know, thou
shalt not mix versions, but just for exploration purposes I thought it
will do me no harm).
To my dismay, the results (using the same input files) look similar, but
(even using the, as far as I can tell, exact same parameters, classical
processing, no mask, same filter, no AR etc.) the significant clusters
are much smaller, some disappear entirely, and the FDR-corrected cutoff
is much higher (2.4 in spm2, 3.49 in spm5). When directly subtracting
the T-maps from each other, the result is almost uniformly negative,
i.e. spm2's T-map has higher T-values throughout the entire brain
(typically on the order of 1.2-2.0).
Now, either I do not see a parameter setting that severely influences my
results in spm5 but not in spm2 or the statistics have become more
strict in some sense. I wonder if there is an explanation for this that
anybody would be willing to share.
Looking forward to your input,
best,
Marko
--
=====================================================================
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
=====================================================================
|