Dear Kately,
Katelyn Norton wrote:
> Hi Will.
> I am so sorry it took me so long to respond, but I really REALLY
> appreciated your response to my question (actually both of them were
> incredibly helpful!!).
>
> The problem I am having with my individual brains is that there is so
> little signal, I am not able to use an FWE correction, let alone an FDR.
> I am having to use uncorrected activation clusters (p<0.005) before I
> see any sort of response (even from the motor cortex which should be
> active in a subject doing a handgrip protocol!).
Yes, this is surprising and suggests there's maybe something more
critically wrong with your analysis. I see that you are including
movement regressors in the model and am wondering if this is removing
your signal. Is it standard to include them in handgrip-fMRI studies ?
Also, you don't have a huge number of events. Only 3 epochs per subject
(or are there multiple sessions ? - did'nt see this in your script). +
there seems an excessive amount of rest (70s) between grip epochs (also
grip epochs are quite long). Its maybe not an optimal use of scanner
time (but then maybe you have other constraints ?)
Other than that I would check each subject in turn. The script you sent
me was for a single subject. You should make sure different subjects
have the appropriate onsets etc. Sorry, its basic stuff I know.
+ am not sure what you mean by a blank design matrix - don't you see the
column for handgrip, and then the columns for the movement regressors ?
BEst,
Will.
So as much as I
> appreciate your support with reporting individual data, you had said at
> the FWE level and I don't know if it's equally as accepted with
> uncorrected volumes? Does this make sense?
>
> Also, I had said that I was having a problem with my group analysis
> which is why I using individuals in the first place. My batches run fine
> through Matlab, but when I select my contrast results to view, there is
> a blank design matrix and no activation/deactivation shows up - even at
> the uncorrected level! So something is obviously wrong... I just don't
> know what and I don't know how to proceed?
>
> I have included both my individual and group Matlab scripts, so perhaps
> you can see a problem/better suggestion?
>
> Again, I am truly very appreciative of your help. If you have the time,
> I would love to hear any suggestions you might have on how to make my
> analysis more efficient.
> Thank you and I look forward to hearing from you soon!
> Katelyn
>
> ----- Original Message -----
> From: Will Penny <[log in to unmask]>
> Date: Monday, March 1, 2010 6:47 am
> Subject: Re: [SPM] Individual Whole Brain Correction
> To: Katelyn Norton <[log in to unmask]>
> Cc: [log in to unmask]
>
> > Hello again,
> >
> > I don't see whats wrong with saying 10/12 subjects showed
> > activation at
> > x,y,z using FWE (p<0.05). Its a precise statement about what
> > you've done so readers should be able to interpret accordingly.
> > I would'nt use different criteria for different subjects.
> >
> > Re the group comparison, you could again take the above
> > approach. But say 4/12 showed activation in group 1 and 8/12 in
> > group 2, all we have is this descriptive statement. And it will
> > take you a fair amount of work to collect together these
> > statements for a number of different brain regions.
> >
> > I think it would be better, and it should be easier to use eg. a
> > two-sample t-test (or ANOVA) at the second level to test for
> > group differences (where in the brain is the group 1 response
> > sig diff from group 2). What problem did you run into with the
> > ANOVA (where in the brain are there any differences between (>2)
> > groups) ?
> >
> > Best,
> >
> > Will.
> >
> > Katelyn Norton wrote:
> > >Hi all.
> > >I am working with two groups of subjects, each having an n=12.
> > After struggling with ANOVA's, I have decided to use whole brain
> > anlyses on each subject and simply report "10/12 showed
> > activation here" for example.
> > >First question, is this an acceptable means of reporting data
> > when I will essentially be comparing these results on a group
> > vs. group basis?
> > >Second question, is it acceptable to use a different level of
> > correction (p value) for each subject or should they all be the
> > same? My hunch is that they should all be the same, however the
> > level of activation among subjects is so variable that for some
> > I can use FWE and for others I cannot correct all (p<0.05).
> > >I would really appreciate any suggestions?
> > >Thanks!
> > >
> > >
> >
> > --
> > William D. Penny
> > Wellcome Trust Centre for Neuroimaging
> > University College London
> > 12 Queen Square
> > London WC1N 3BG
> >
> > Tel: 020 7833 7475
> > FAX: 020 7813 1420
> > Email: [log in to unmask]
> > URL: http://www.fil.ion.ucl.ac.uk/~wpenny/
> >
--
William D. Penny
Wellcome Trust Centre for Neuroimaging
University College London
12 Queen Square
London WC1N 3BG
Tel: 020 7833 7475
FAX: 020 7813 1420
Email: [log in to unmask]
URL: http://www.fil.ion.ucl.ac.uk/~wpenny/
|