headers
Matthew Brett | 1 Sep 10:03
Picon
Gravatar

Re: % signal change clarifications

Hi,

> I've red a lot of documentation about the % signal change calculation for a
> ROI but I'm not sure to understand. I proceed as follow for each subject:
> After running the ROI first level analysis in marsbar via GUI select Results
> -> % signal change -> "New" for Event type -> Add the event of interest ->
> "0" for event duration (I've an event related design) -> I take the OUTPUT
> VALUE in the command window

So - that output value is the marsbar calculation of the percent
signal change from this event.  This will have taken into account
multiple basis functions, and the intercept that you mention below.

-> Results -> MarsBar SPM graph -> Contrast
> estimates and 90% C.l. -> select Effects of interest -> in the command
> window I type beta -> I take the INTERCEPT (last column) -> %signal change =
> (OUTPUT VALUE/INTERCEPT)*100.

No - so the output value above has already been divided by the
intercept, so this step is not correct.

> If all is ok, Can I average the % signal change for each subject to obtain a
> mean % signal change for the group?

Yes, that's quite a common thing to do.

> Can I take different % signal change if I include or don't include the 6
> realign regressors in the design matrix?

You will get slightly or even very different answers with the realign
(Continue reading)

Permalink | Reply | 0 comments |
headers
Matthew Brett | 1 Sep 10:08
Picon
Gravatar

Re: specifications about ROI

Hi,

> After running a roi analysis with marsbar (structural rois based on aal)
> I've a corrected p-value for each ROIs. I need some advices about what I
> must report together with this values for example the MNI coordinates for
> each region or the volume for each region or the number of voxels for each
> region or what else? Which are the necessary information for a publication?
> Moreover neuroimaging papers usually reports t/z/F-maps so my question is:
> what kind of activation map shall I report when I perform a roi analysis?

Sorry to be a bit slow to get back to you.  I think you only have to
report the names of the structural ROIs you used together with the
paper they were based on (it's referenced on the marsbar site) and the
link to the download.  You could probably also report the volume, or
even provide an image of the ROI displayed on a template brain - if
you have space.

For t/F maps - I personally prefer to see the whole brain activation
map as well as the ROI analyses.

I hope that answers your question - let me know if not,

Best,

Matthew

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
(Continue reading)

Permalink | Reply | 0 comments |
headers
Federico Tubaldi | 1 Sep 15:16
Picon
Favicon

Set FIR model


Dear Matthew and dear All,

I like set a FIR model for my ROI analysis as Matthew suggested me some days
ago. I've red a lot of basic documentation about FIR model on spm and marbar
lists. Thanks to the Rik and Matthew e-correspondence on spm list I've
understood the importance to set suitable parameters.
Very briefly: I've a jittered randomized event related design, event
duration (visual stimulus) = 2 sec, TR = 3 sec, the investigated cognitive
process associated to the event observation is complex (related to social
cognition and theory of mind), the roi is the anterior cingulate.

My first question is: Is this actually an event related design that is shall
I set DURATION = 0 sec or it's better I set DURATION = 2 sec?

Which values for "window lenght" and "order" of the FIR model you suggest me
considering the complexity of the investigated cognitive process (I think
that the neural activity induced by the event could be sustained and
prolonged)?

Thank you.   

--

-- 
No virus found in this outgoing message.
Checked by AVG Free Edition.
Version: 7.1.405 / Virus Database: 268.11.7/435 - Release Date: 31/08/2006

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
(Continue reading)

Permalink | Reply | 1 comment |
headers
Matthew Brett | 1 Sep 15:42
Picon
Gravatar

Re: Set FIR model

Hi,

> My first question is: Is this actually an event related design that is shall
> I set DURATION = 0 sec or it's better I set DURATION = 2 sec?

Personally I would use the 2 second duration; the model fit will not
be _very_ different though.

> Which values for "window lenght" and "order" of the FIR model you suggest me
> considering the complexity of the investigated cognitive process (I think
> that the neural activity induced by the event could be sustained and
> prolonged)?

Shorter window length and higher 'order' (number of bins) will mean
more noise in your estimates.  There will be a trade-off between noise
and the time details and duration (order) of your time courses.  Why
not start with the bin length of 1TR and - say - about 27 seconds for
your HRF to play out - meaning an order of 9...

Best,

Matthew

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
Permalink | Reply | 0 comments |
headers
Federico Tubaldi | 2 Sep 12:24
Picon
Favicon

Extract whole brain PSTH values

Dear Matthew and dear All,

I'd like use MarsBar rather than Spm 2 to run a whole brain analysis because
I want to extract PSTH values for each subject at a precise maxima voxel (in
SPM I can't) so I follow these steps in MarsBar: Design -> FMRI Models ->"I
define my model" -> Design -> Add images to FMRI design -> Remove Global
effects -> "None" -> Design -> Add/edit filter for FMRI design -> "128" ->
Correct for serial correlation -> "fmristat AR(2)" -> Results -> Estimate
results -> MarsBar ask me SELECT ROI DATA (*_mdata.mat) BUT I DON'T WANT TO
DO A ROI ANALYSIS. How can I continue to estimate my whole brain model?

Also I try: Design -> Set design from file -> "I select SPM.mat (design
estimated in spm) -> Results -> Statistc table -> I DON'T SEE THE CONTRAST
MANAGER but MarsBar ask me SELECT MARSBAR ESTIMATED DESIGN (*_mres.mat).  

Thank you.

--

-- 
No virus found in this outgoing message.
Checked by AVG Free Edition.
Version: 7.1.405 / Virus Database: 268.11.7/436 - Release Date: 01/09/2006

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
Permalink | Reply | 0 comments |
headers
Carolyn L. Fort | 10 Sep 23:17
Picon

signal intensity plots

Hello.

When plotting the extracted data from an ROI (this is under the Data menu, option Plot Data (simple), NOT in
the Results menu, option Marsbar SPM Graph) I get a plot of signal intensity across the timecourse of my
experiment. Are these values adjusted in any way, eg, for global flow? Also, once I've saved this plot and
am working with the resultant *_mdata.mat file, what are the values stored in "y" and "Yvar" in the
workspace? It seems logical that Yvar would be variance, but the values are too high to make sense. I'm
having a hard time reconciling these files of similar names to those produced when using the Marsbar SPM
Graph option (Results menu).

Many thanks,

Carolyn

Carolyn L. Fort
Research Specialist, Sr.
Department of Psychiatry
University of Arizona
PO Box 245002
Tucson, AZ 85724-5002
v: 520-626-8568
f: 520-626-6050

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
Permalink | Reply | 1 comment |
headers
Carolyn L. Fort | 10 Sep 23:29
Picon

signal intensity within a contrast

I've been looking at plots of signal intensity within an ROI across an entire timecourse, but is there a way
to obtain a measure of signal intensity for a particular contrast?

Thanks,

Carolyn

Carolyn L. Fort
Research Specialist, Sr.
Department of Psychiatry
University of Arizona
PO Box 245002
Tucson, AZ 85724-5002
v: 520-626-8568
f: 520-626-6050

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
Permalink | Reply | 1 comment |
headers
Matthew Brett | 11 Sep 08:28
Picon
Gravatar

Re: signal intensity within a contrast

Hi,

On 9/10/06, Carolyn L. Fort <clfort <at> mac.com> wrote:
> I've been looking at plots of signal intensity within an ROI across an entire timecourse, but is there a way
to obtain a measure of signal intensity for a particular contrast?

Yes - you can either:

Plot the event related time course using the event related options or

Plot the values your contrast using the marsbar spm graph option or

Extract the betas or contrast values using the batch system - see the
marsbar faq...

Best,

Matthew

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
Permalink | Reply | 0 comments |
headers
Matthew Brett | 11 Sep 08:36
Picon
Gravatar

Re: signal intensity plots

Hi,

> When plotting the extracted data from an ROI (this is under the Data menu, option Plot Data (simple), NOT in
the Results menu, option Marsbar SPM Graph) I get a plot of signal intensity across the timecourse of my
experiment. Are these values adjusted in any way, eg, for global flow?

Yes - the SPM scaling has been applied, if you got these images via
the SPM design.  The adjustment for FMRI data is very simple, and
merely normalizes the grand mean signal across the brain (see
http://imaging.mrc-cbu.cam.ac.uk/imaging/PrinciplesStatistics#global_signal
and
http://imaging.mrc-cbu.cam.ac.uk/imaging/PrinciplesStatistics#grand_mean)
 to be 100 for each session.

> Also, once I've saved this plot and am working with the resultant *_mdata.mat file, what are the values
stored in "y" and "Yvar" in the workspace? It seems logical that Yvar would be variance, but the values are
too high to make sense. I'm having a hard time reconciling these files of similar names to those produced
when using the Marsbar SPM Graph option (Results menu).

y is the data summarized for the ROI - usually the mean signal across
the ROI therefore.  Yvar is the variance across the ROI,

Best,

Matthew

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
(Continue reading)

Permalink | Reply | 0 comments |
headers
Nicole Dudukovic | 11 Sep 21:14
Picon

FIR and parametric modulation


Hi,

I have a design matrix with three trial types.  To account for 
within-subjects RT differences between tasks, I've added in a fourth 
trial type that includes all 3 conditions and is parametrically 
modulated with trial specific RT.  When I take ROIs defined from my 
SPM contrasts and use the marsbar batch script to obtain FIR 
timecourses, the timecourses don't seem to match the SPM stats.  The 
condition with the most variance in RT always seems have the greatest 
percent signal change.  Could the parametric modulation be 
interfering with the deconvolution?  None of the trial types that I'm 
extracting have been modulated, and I can't figure out why the 
timecourses don't match.  I don't have this problem when I model the 
experiment without the 4th trial type and without parametric 
modulations.

Any help would be appreciated!

Thanks,
Nicole

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
Permalink | Reply | 0 comments |

Gmane