 |
 |
I think Vince's logic for combining the canonical HRF parameter estimate with those of its derivatives extends naturally to both temporal and dispersion derivatives (or indeed any other first-order partial derivative). However, one thing you might want to be aware of is that the distribution of the quantity sign(V1).*sqrt(V1.^2+V2.^2) is not Gaussian: the second (sqrt) term will always be positive (even if noise alone), which will result in a bimodal distribution around zero under the null hypothesis, which may caution against parametric statistics. Vince may have explored this in his paper (apologies for being too lazy to look!), and may want to comment, but you could consider using SnPM in this case. BW,R -----Original Message----- From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On Behalf Of Chris Watson Sent: 13 March 2012 19:12 To: [log in to unmask] Subject: Re: [SPM] HRF + derivatives From the email you linked to: you would first obtain the individual contrasts for each of the canonical and derivative terms from each subject (two contrast volumes per subject), as before. Then, instead of entering these volumes directly into a second level analysis, you would compute, for each subject, a single volume estimating the "amplitude" of the effects at each voxel = sign(V1).*sqrt(V1.^2+V2.^2), where V1 is the canonical effect contrast volume, and V2 is the temporal derivative effect contrast volume. These "amplitude" effects estimate the amplitude of the peak response irrespective of the delay at which it occurrs (within a reasonable range). Last you would enter these "amplitude" volumes in a simple second-level t-test for population inferences. On 03/13/2012 02:07 PM, Laura Tully wrote: > Thanks Jonathan, that does help. I'm now trying to work out how to > define the appropriate t-contrasts in order to compute the "amplitude" > at each voxel as described by Sue Gabriele here (and discussed in > Calhoun et al. 2004): > https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=SPM;7fba87b3.0811 > > I think that I need to define t-contrasts at the individual level for > each of the three basis functions (canonical, time, dispersion) for > all of my conditions [e.g. 1 0 0; 0 1 0; 0 0 1] Is that correct? The > bit that I get stuck on is what to do next... according to Calhoun et > al (2004) it looks like they created paired difference maps between > conditions as well [CondA(allterms)-CondA(derivatives)] - > [condB(allterms)-CondB(derivatives)] but I'm not quite sure how to do > this, or how it relates to the "amplitude" computation that is > discussed in the paper. Any light you could shed on this issue would > be most appreciated! > > Laura. > > On Tue, Mar 13, 2012 at 12:45 PM, Jonathan Peelle <[log in to unmask] > <mailto:[log in to unmask]>> wrote: > > Dear Laura, > > > could someone clarify for me what the betas produced using time and > > dispersion derivatives are? Is it that the first is canonical > only, the > > second is canonical+time, and the third is > canonical+time+dispersion, OR is > > it canonical only, time only, and dispersion only? > > It's the latter-canonical only, time only, and dispersion only. When > you estimate a model in SPM, the beta reflects the contribution of > that model in your design matrix: beta_0001 is the first column, > beta_0002 the second column, etc. > > Hope this helps! > > Best regards, > > Jonathan > > -- > Dr. Jonathan Peelle > Center for Cognitive Neuroscience and > Department of Neurology > University of Pennsylvania > 3 West Gates > 3400 Spruce Street > Philadelphia, PA 19104 > USA > http://jonathanpeelle.net/ > > > > > -- > Laura Tully > Social Neuroscience & Psychopathology > Harvard University > 840 William James Hall > 33 Kirkland St > Cambridge, MA 02138 > [log in to unmask] <mailto:[log in to unmask]> > >