Regarding the scanner starting:
Friston had a post on this a long time ago,
http://www.jiscmail.ac.uk/cgi-bin/wa.exe?A2=SPM;6f09b9d2.00
though of course I'm sure lots of us have thought about this issue. Because of my background I tend to think of it in terms of convolution with a constant not giving back a constant (unless as you point out the fixation/baseline starts before the scanner started). How large do you think this effect is? It's only transient, but my impression is that it can't be disregarded in many cases. People don't seem to talk about it much.
About gPPI vs the "subtraction method": are you saying there's a difference even in the simplest, canonical PPI examples?
Best,
Stephen J. Fromm, PhD
Contractor, NIMH/MAP
(301) 451--9265
________________________________
From: MCLAREN, Donald [[log in to unmask]]
Sent: Tuesday, September 27, 2011 4:37 PM
To: Fromm, Stephen (NIH/NIMH) [C]
Cc: [log in to unmask]
Subject: Re: [SPM] PPI using fixation: model explicitly
The conclusion:
If its truly fixation, then either a single condition or gPPI can be used. You cannot use the subtraction method because its a poor model fit and has dramatically different values.
If its fixation and evokes a response, then its really 2 conditions and gPPI should be used.
Best Regards, Donald McLaren
=================
D.G. McLaren, Ph.D.
Postdoctoral Research Fellow, GRECC, Bedford VA
Research Fellow, Department of Neurology, Massachusetts General Hospital and
Harvard Medical School
Office: (773) 406-2464
=====================
This e-mail contains CONFIDENTIAL INFORMATION which may contain PROTECTED
HEALTHCARE INFORMATION and may also be LEGALLY PRIVILEGED and which is
intended only for the use of the individual or entity named above. If the
reader of the e-mail is not the intended recipient or the employee or agent
responsible for delivering it to the intended recipient, you are hereby
notified that you are in possession of confidential and privileged
information. Any unauthorized use, disclosure, copying or the taking of any
action in reliance on the contents of this information is strictly
prohibited and may be unlawful. If you have received this e-mail
unintentionally, please immediately notify the sender via telephone at (773)
406-2464 or email.
On Tue, Sep 27, 2011 at 4:35 PM, MCLAREN, Donald <[log in to unmask]<mailto:[log in to unmask]>> wrote:
I just finished running some simulations on this topic. The results and thoughts are as follows:
(1) By definition, there should be no change in the BOLD signal during the viewing of the implicit baseline or fixation periods. Ideally, the baseline/fixation began several seconds before the scanner started. If the scanner starts, then you bring up the fixation cross, one might argue that the fixation cross will evoke a BOLD response because you've changed the stimulus that the subject is exposed to after the scan starts.
(2) Now, we either have fixation evoking no response or we have fixation/baseline evoking a response because it started after the scanner started.
(3) If fixation does not evoke a BOLD response and does not change the connectivity (e.g. connectivity was there before the scan started), then the only source of difference between comparing task versus fixation/implicit baseline and comparing task versus explicit fixation/baseline is random noise. This results in two possibilities:
(a) minor differences over 100000 (100 PPIs each with 100 voxels) tests if gPPI is used with 2 conditions (B=0.9889,SS=3.5447 -gPPI; 0.9876,SS=3.6676 -SPM; correlation=.52)
(b) bigger differences over 100000 tests if SPMs subtraction is used (B=0.8675,SS=6.4581 -SPM subtract 2 conditions; 0.9876,SS=3.6676 -SPM -SPM 1 condition;correlation=.48)
Importantly, all models used the same random noise and same connectivity differences. Thus, the only difference is the model.
SS --> Sum of Squares
The difference between gPPI and SPM subtraction is that gPPI models the entire experimental space, whereas SPMs subtraction method only models the explicit comparison being made in the analysis (e.g. condition1-condition2 and ignoring all other conditions).
Hopefully, these simulations are useful in your modelling of your data.
Best Regards, Donald McLaren
=================
D.G. McLaren, Ph.D.
Postdoctoral Research Fellow, GRECC, Bedford VA
Research Fellow, Department of Neurology, Massachusetts General Hospital and
Harvard Medical School
Office: (773) 406-2464<tel:%28773%29%20406-2464>
=====================
This e-mail contains CONFIDENTIAL INFORMATION which may contain PROTECTED
HEALTHCARE INFORMATION and may also be LEGALLY PRIVILEGED and which is
intended only for the use of the individual or entity named above. If the
reader of the e-mail is not the intended recipient or the employee or agent
responsible for delivering it to the intended recipient, you are hereby
notified that you are in possession of confidential and privileged
information. Any unauthorized use, disclosure, copying or the taking of any
action in reliance on the contents of this information is strictly
prohibited and may be unlawful. If you have received this e-mail
unintentionally, please immediately notify the sender via telephone at (773)
406-2464 or email.
On Sat, Sep 24, 2011 at 11:22 AM, Stephen J. Fromm <[log in to unmask]<mailto:[log in to unmask]>> wrote:
Occasionally I run across experimenters who express a desire to run a PPI, where the psych variable for the PPI is of the form "condition X vs fixation" or "condition X vs baseline."
In a thread from a few years ago, Darren G. claimed that to do this, one must model fixation explicitly in the conventional model that feeds into PPI. Donald (McLaren) has probably thought about this, but I haven't had time to read his code/poster.
Any thoughts? (Explicitly modeling fixation is laborious because it means all the subject-level models must be redesigned.)
|