Hi Kelly,
what we have been doing is to estimate the region-specific true hrf using an
FIR model (20 seconds, 10 time bins) on main effects in an F-test. This
works best if you do a 'long-trial' version of your experiment, with large
ISIs, but it actually also works on 'regularly' timed ER experiments, as
long as you use plenty of jitter (we have compared the two). So, for each
region of interest, we end up with a main-effects hr slope. Then, in ROI
analyses, we use t-tests weighting the FIR time bins with these values, to
compare conditions, just like with regular contrasts (e.g. 1 -1), but with
10 values for each condition, instead of 1 or 2.
This seems to work quite well, but the big downside, of course, is that one
can only do it region-specifically. Also, the models tend to end up pretty
big, if you have a lot of contrasts/regions of interest. It is true, though,
that the true hrf can deviate from the canonical to such an extent that a
canonical-hrf analysis (even with time derivatives) will simply not pick up
activation that might well be perfectly valid.
Hope this makes sense,
Dirk
************************************
Dirk-Bart den Ouden, Ph.D.
Aphasia & Neurolinguistics Research Laboratory
Dept. of Communication Sciences & Disorders
Northwestern University
2240 Campus Drive, Evanston IL 60208-3066
Phone: (office) 1-847-467-2515; (lab) 1-847-467-7591
************************************
%%-----Original Message-----
%%From: SPM (Statistical Parametric Mapping)
%%[mailto:[log in to unmask]] On Behalf Of Kelly Westlake
%%Sent: Thursday, May 14, 2009 11:03 AM
%%To: [log in to unmask]
%%Subject: Re: [SPM] Which basis function?
%%
%%Hi Mike,
%%
%%Thanks for the response. I tried using the HRF with
%%derivatives and it seems to increase the signal a bit. Just
%%for the sake of comparison, have you tried using the FIR or
%%any of the other functions? Since it may not be appropriate
%%for me to assume a canonical HRF even with derivatives in my
%%subject population, I'd like to try other functions, but I'm
%%not sure what to include as my window length or order. Do you
%%know what these inputs should be?
%%
%%Best,
%%Kelly
%%
|