HI - we will try to get this added in for future versions,
Cheers.
On 21 Apr 2009, at 11:55, Kern, Kyle wrote:
> Thank you Tom. Would you predict that future versions of randomise
> will be able to accomodate this type of design? I imagine it is a
> pretty common setup for longitudinal studies. Also, can anyone
> suggest alternative approaches to determining the group x time
> interaction for tbss/vbm? This will be an important issue for me in
> an upcoming study comparing treatment effects.
>
> thanks a lot,
>
> Kyle Kern
>
>
>
>
> ________________________________________
> From: FSL - FMRIB's Software Library [[log in to unmask]] On Behalf
> Of Thomas Nichols [[log in to unmask]]
> Sent: Tuesday, April 21, 2009 3:09 AM
> To: [log in to unmask]
> Subject: Re: [FSL] 2x4 ANOVA, 1 repeated measure
>
> Dear Kyle,
>
> Sorry for the delay. The reason that no one has chimed in is that
> randomise can't deal with this design in general. As you concisely
> state, you have 2-way repeated measures ANOVA, 1 within-subject
> factor (time) and 1 between subject factor (group) (and the repeated
> 'cluster' variable subject).
>
> In this advanced model, the appropriate way to do a permutation test
> depends on the question asked. If you are asking just about the
> within-subject factor (time), randomise *could* deal with this, just
> use the model as you have specified it *but* be sure to specify an
> exchangeability block file marking the subjects (e.g., transposed
> it would like like
> 1 1 1 1 2 2 2 2 3 3 3 3 ...
> ). Randomise will then know to only permute scans within subject
> (appropriate for a null hypothesis of no time effect, *and*
> implicitly assuming exchangeability within subject).
>
> But if you're asking about the between-subject factor, the
> appropriate way to permute is to shuffle subjects' data as a whole
> around. Or, equivalently to, permute the design ensuring that only
> permutations that move around sets of four rows together are used).
> That currently isn't implemented in randomise.
>
> Hope this helped!
>
> -Tom
>
>
> On Wed, Mar 25, 2009 at 12:34 AM, Kern, Kyle <[log in to unmask]<mailto:[log in to unmask]
> >> wrote:
> Hi fsl users,
>
> I have a tbss dataset with a control group, a patient group, and 4
> timepoints for each subject. I am trying to create a GLM design that
> essentially models a 2x4 ANOVA with 1 repeated measure (time). I was
> hoping that anyone who has worked with a similar design could look
> over my setup and steer me in the right direction. This study has
> 11 controls and 21 patients, but for simplicity, the following
> design has 3 controls and 4 patients. I would appreciate any
> corrections or better approaches to the design below:
>
> .mat file:
>
> cont1_T1 1 0 0 0 0 0 1 0 0 0 0 0 0
> cont1_T2 0 1 0 0 0 0 1 0 0 0 0 0 0
> cont1_T3 0 0 1 0 0 0 1 0 0 0 0 0 0
> cont1_T4 0 0 0 0 0 0 1 0 0 0 0 0 0
> cont2_T1 1 0 0 0 0 0 0 1 0 0 0 0 0
> cont2_T2 0 1 0 0 0 0 0 1 0 0 0 0 0
> cont2_T3 0 0 1 0 0 0 0 1 0 0 0 0 0
> cont2_T4 0 0 0 0 0 0 0 1 0 0 0 0 0
> cont3_T1 1 0 0 0 0 0 0 0 1 0 0 0 0
> cont3_T2 0 1 0 0 0 0 0 0 1 0 0 0 0
> cont3_T3 0 0 1 0 0 0 0 0 1 0 0 0 0
> cont3_T4 0 0 0 0 0 0 0 0 1 0 0 0 0
> pat1_T1 0 0 0 1 0 0 0 0 0 1 0 0 0
> pat1_T2 0 0 0 0 1 0 0 0 0 1 0 0 0
> pat1_T3 0 0 0 0 0 1 0 0 0 1 0 0 0
> pat1_T4 0 0 0 0 0 0 0 0 0 1 0 0 0
> pat2_T1 0 0 0 1 0 0 0 0 0 0 1 0 0
> pat2_T2 0 0 0 0 1 0 0 0 0 0 1 0 0
> pat2_T3 0 0 0 0 0 1 0 0 0 0 1 0 0
> pat2_T4 0 0 0 0 0 0 0 0 0 0 1 0 0
> pat3_T1 0 0 0 1 0 0 0 0 0 0 0 1 0
> pat3_T2 0 0 0 0 1 0 0 0 0 0 0 1 0
> pat3_T3 0 0 0 0 0 1 0 0 0 0 0 1 0
> pat3_T4 0 0 0 0 0 0 0 0 0 0 0 1 0
> pat4_T1 0 0 0 1 0 0 0 0 0 0 0 0 1
> pat4_T2 0 0 0 0 1 0 0 0 0 0 0 0 1
> pat4_T3 0 0 0 0 0 1 0 0 0 0 0 0 1
> pat4_T4 0 0 0 0 0 0 0 0 0 0 0 0 1
>
> EVs 1-3 model controls at each timepoint, 4-6 model patients at each
> timepoint, and 7-13 model the individual subject means.
>
> contrasts:
>
> Controls-Patients: 1 1 1 -1 -1 -1 0 0 0 0 0 0 0
> X Ftest1: Main effect of Group
> Time1-Time4: 1 0 0 1 0 0 0 0 0 0 0 0
> 0 X
> Time2-Time4: 0 1 0 0 1 0 0 0 0 0 0 0
> 0 X Ftest2: Main effect of Time
> Time3-Time4: 0 0 1 0 0 1 0 0 0 0 0 0
> 0 X
> C(T1-T4)-P(T1-T4): 1 0 0 -1 0 0 0 0 0 0 0 0
> 0 X
> C(T2-T4)-P(T2-T4): 0 1 0 0 -1 0 0 0 0 0 0 0
> 0 X Ftest3: Interaction Group x Time
> C(T3-T4)-P(T3-T4): 0 0 1 0 0 -1 0 0 0 0 0 0
> 0 X
>
> And I can also include contrasts for T1-T3, T2-T3, T1-T2, C(T1-T3)-
> P(T1-T3), C(T2-T3)-P(T2-T3), C(T1-T2)-P(T1-T2), but not include them
> as ftests right? (since they are linear combinations of the others)
> I am not sure how the data should be grouped for the exchangability
> blocks, so I have them all in one group in the .grp file, is that
> correct?
>
> Does this setup look appropriate?
> attached are the graphical representations.
>
> thanks for the help,
>
> Kyle Kern
>
>
>
>
> ________________________________________
> From: FSL - FMRIB's Software Library [[log in to unmask]<mailto:[log in to unmask]
> >] On Behalf Of MCLAREN, Donald [[log in to unmask]<mailto:[log in to unmask]
> >]
> Sent: Wednesday, March 18, 2009 10:06 PM
> To: [log in to unmask]<mailto:[log in to unmask]>
> Subject: Re: [FSL] modeling slope over time
>
> Instead of estimating the within-subject slopes, I'd do a repeated
> measures ANOVA. In the ANOVA, you will want subject, group, time,
> and a group-by-time interaction term. The reason for doing it this
> way is to avoid the assumption that the change is linear with time.
>
> On Wed, Mar 18, 2009 at 10:01 PM, Kyle Kern <[log in to unmask]<mailto:[log in to unmask]
> ><mailto:[log in to unmask]<mailto:[log in to unmask]>>>
> wrote:
> Hi,
> I am trying to set up a longitudinal tbss analysis over 4 time
> points contrasting a patient
> group to a control group. I'm interested in change over time rather
> than the effect of any
> specifc time point. I was planning to model the slope for each
> subject and then put this
> into a group t-test. Would modelling the slope for each subject be
> the same approach as a
> simple correlation where my design matrix might look like this?:
>
> -3
> -1
> 1
> 3
>
> Would this reflect the magnitude of the slope or does it only
> reflect the linearity like a
> correlation coefficient? I would hypothesize that patients show a
> steeper slope than
> controls though they may show more variability, so a least squares
> regression slope
> might detect differences but a simple correlation coefficient might
> not.
>
> Also, to perform a 2 level analysis with tbss is it preferable to
> use randomise and feed
> the within subject t-stat into the group comparison or should I use
> the feat gui and treat
> each within subject analysis as a timeseries and feed that into a
> higher level mixed
> effects analysis?
>
> thanks for all the help,
>
> Kyle Kern
>
>
>
> --
> Best Regards, Donald McLaren
> =====================
> D.G. McLaren
> University of Wisconsin - Madison
> Neuroscience Training Program
> Office: (608) 265-9672
> Lab: (608) 256-1901 ext 12914
> =====================
> 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 (608) 265-9672 or email.
>
> IMPORTANT WARNING: This email (and any attachments) is only
> intended for the use of the person or entity to which it is
> addressed, and may contain information that is privileged and
> confidential. You, the recipient, are obligated to maintain it in a
> safe, secure and confidential manner. Unauthorized redisclosure or
> failure to maintain confidentiality may subject you to federal and
> state penalties. If you are not the intended recipient, please
> immediately notify us by return email, and delete this message from
> your computer.
>
>
>
> --
> ____________________________________________
> Thomas Nichols, PhD
> Director, Modelling & Genetics
> GlaxoSmithKline Clinical Imaging Centre
>
> Senior Research Fellow
> Oxford University FMRIB Centre
>
> IMPORTANT WARNING: This email (and any attachments) is only
> intended for the use of the person or entity to which it is
> addressed, and may contain information that is privileged and
> confidential. You, the recipient, are obligated to maintain it in a
> safe, secure and confidential manner. Unauthorized redisclosure or
> failure to maintain confidentiality may subject you to federal and
> state penalties. If you are not the intended recipient, please
> immediately notify us by return email, and delete this message from
> your computer.
>
---------------------------------------------------------------------------
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director, Oxford University FMRIB Centre
FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
---------------------------------------------------------------------------
|
