You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Similarly to what we have done in the SPM NIDM viewer, could we recompute Z from P when Z has an infinite value : cf. incf-nidash/nidmresults-spmhtml#39?
snpm_t2z.m was inherited from SPM99 which used to Gaussianise everything. I’m pretty sure that I later compared the accuracy of it to just
norminv(tcdf(x),df))
for x<0 and
-norminv(tcdf(-x),df))
for x>= 0
and found the function was no better than this (or, equivalently, with the SPM functions).
The issue is that, in spm_list, instead of waiting for underflow to occur, there is a check against a given tolerance level and if the pvalue is smaller than it you get Inf
Similarly to what we have done in the SPM NIDM viewer, could we recompute Z from P when Z has an infinite value : cf. incf-nidash/nidmresults-spmhtml#39?
@gllmflndn: the Inf are initially created by SPM at https://github.com/spm/spm12/blob/master/spm_list.m#L464-L470, what do you think of this proposed update (in the NIDM exporter or in SPM?)?
The text was updated successfully, but these errors were encountered: