headers
Marina Papoutsi | 2 Oct 17:41
Picon
Picon

High F-contrast and ESS values


Dear Marsbar-users

I have a question wrt running an FIR RFX analyis in marsbar.
For my analysis, I have modeled my events using an FIR with 16 bins and 
length=TR (2s).
I have performed a second level analysis in SPM5 and would like to also 
perform an ROI analysis on one of the regions.
I have 14 subjects * 16 bf = 224 timepoints and 1 region. I run a 
first-level ROI analysis and extract summary values for each of these 
points. I then write them out to a matrix (size: 224x1) to use as the 
data for a second level analysis.

I load the SPM second level design matrix and import the matrix with the 
timepoints as the data (option summary time course for region). I then  
estimate the model, provide a basic F-contrast (eye(number of basis 
functions)) and look at the statistics table below:

Contrast name                   ROI name:       Extra SS:    F 
statistic:  Uncorrected P:    Corrected P  
--------------------------------------------------------------------------------------------------------------------------------------------------

 
Fcon-all                       Matlab input_1:   242872823.12:   
697149249.02:       0.000000:       0.000000

For obvious reasons, I think that these results look really suspicious. 
But I am not sure what is wrong.
Am I doing something wrong or is is a problem with non-sphericity?
(Continue reading)

Permalink | Reply | 0 comments |
headers
Jay Hegdé | 4 Oct 07:13
Picon

Probelms with compute_contrasts() ?

Dear MarsBaR experts,
    I'm trying to model an ROI using a batch script (adapted from the 
code provided in the marsbar FAQ).  It works without a hitch (using 
marsbar-devel in matlab 7.0.4), but I don't understand the output of 
compute_contrasts(), specifically MVres (please see below).

    I just need a way of visualizing both the actual and modeled time 
series (with or without compute_contrasts).  How do I do this?

    Thanks in advance,
    Jay

------------------------------------------------------------------------------------
spm_name = '/my/path/SPM.mat';

roi_file = '/my/path/my_roi.mat';

% Make marsbar design object
D  = mardo(spm_name);

% Make marsbar ROI object
R  = maroi(roi_file);

% Fetch data into marsbar data object
Y  = get_marsy(R, D, 'mean');

x = summary_data(Y); % get summary time course(s)
plot(x, 'r-');      %<<<< Looks fine.
......
(Continue reading)

Permalink | Reply | 1 comment |
headers
Marina Papoutsi | 16 Oct 12:35
Picon
Picon

Estimating covariance using AR(2) at the second-level (RFX): please help!


Dear marsbar-users and Dr. Brett,

Following my last post 
(http://thread.gmane.org/gmane.comp.graphics.spm.marsbar/631), I have a 
few more questions, which I hope that someone will be able to answer for 
me.

I have been trying to run a second-level ROI analysis using marsbar. My 
second-level model includes one ROI region, one condition and 16 basis 
functions. To estimate the second-level model, I extract the con values 
from the single-subject analysis matrices and import them into marsbar. 
I also load the second-level design that I have previously estimated 
with SPM (for a whole-brain FIR analysis). Estimation seems to be 
without any problems, but when I look at the results, the contrast 
values are huge (in the range of millions). I have looked around a bit 
and it seems that most of the problems are generated due to issues with 
the covariance matrix estimation.

My questions:
1) Can I use AR to estimate the covariance? It seems that this option is 
not available for second-level models. I am not sure why and I was 
wondering whether someone could explain this to me.
2) Why does the covariance matrix estimation fail? What are the 
implications of not redoing the estimation of the covariance and 
whitening matrix and simply using the SPM originally estimated matrix 
(for the whole brain?). I understand that the whole-brain estimated 
matrix is also based on an average time-course and then generalized. 
Even though this maybe not ideal, I was wondering whether it would 
suffice in my case, since I am having all these problems.
(Continue reading)

Permalink | Reply | 0 comments |
headers
Ana Raposo | 17 Oct 16:07
Favicon

extracting FIR timecourses

Hi all,

I am trying to extract FIR timecourses in a study that uses a sparse-imaging
design with a TR=10.2s. The first 8.6s of the TR are a silent period (during
which a spoken sentence was presented) and the last 1.6s correspond to the scan
acquisition. Is it possible to extract timecourses in a sparse-imaging
design with such TR?

When I try the script in
http://marsbar.sourceforge.net/faq.html#extract_fir

I get the following error message:

??? Error using ==> event_fitted_fir
Too many input arguments.

Error in ==> /work/imaging/Functional/KTB/ffx/marsbar_batch.m
On line 95 ==>   fir_tc(:, e_s) = event_fitted_fir(E, e_specs(:,e_s), bin_size,
...

Any suggestions of what may be causing the error and how to fix it would be
great.

Thanks in advance!
Ana

..............................
Ana Raposo, PhD
Psychology and Neuroscience
Duke University
(Continue reading)

Permalink | Reply | 1 comment |
headers
Jérôme Redouté | 18 Oct 15:35
Picon

Set design from SPM5 SPM.mat

Dear all,
I unsuccessfully try to set my design in Marsbar from a previously 
estimated design in SPM5.
I get this warning message: "this does not appear to be a valid design".
I suspected broken links with original images files, but if I try with 
another SPM.mat from another experiment, estimated in another computer, 
but with SPM2, it works perfectly well.
How can I fix that? is there something to do to make Marsbar working 
with SPM5?
Thanks for your help
Jerome

--

-- 
---------------------------------
Jerome Redouté, Ph. D.
Reward and decision making team
Centre de Neuroscience Cognitive
CNRS-UMR 5229 - UCBL
67, Bd. Pinel. 69675 Bron
Tel: 04 37 91 12 44
jerome.redoute <at> isc.cnrs.fr
---------------------------------

-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/
Permalink | Reply | 0 comments |
headers
jayhegde | 23 Oct 04:20
Picon

Adjusting for effects of interest in batch mode


Hi,

  I know one can adjust for effects of interest using the MarsBaR GUI.  But
is it possible to do this in batch mode (i.e., using a script)?

  Thank you,
  Jay

--

-- 
View this message in context: http://www.nabble.com/data-extraction-from-MarsBar-tf1712188.html#a13356313
Sent from the MarsBaR mailing list archive at Nabble.com.

-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/
Permalink | Reply | 0 comments |
headers
Michiru Makuuchi | 23 Oct 14:16
Picon
Favicon

event time course with FIR model, option 'single'

Hi,

I have plotted event time course by FIR model with satisfactory results.
However, I am not sure if I set the model correctly.

The experiment has 10 conditions, namely A,B,C,D,E,qA,qB,qC,qD,and qE.  
Condition A to E have a duration of 6 s presenting word sequences. qA to qE
have a duration of 9 s showing the word sequences and comprehension questions.
Our main intrest is on the condions of A to E, and they occupy the 80 %
of trials. 

My concern is that I chose 'single' but not 'stacked.
Are the plots for condition A to E valid?

Thank you in advance.

Michiru

Dr. Michiru Makuuchi 
Max Planck Institute for Human Cognitive and Brain Sciences 
Stephanstrasse 1a, 04103 Leipzig, Germany 

-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/
Permalink | Reply | 0 comments |
headers
Jana Schaich Borg 02 | 28 Oct 01:29
Favicon

reconstructing percent signal change from timecourse

Hi,

  We are trying out a new analysis and want to do a bunch of internal checks as we go along.  We have extracted the
adjusted timecourse for each of our subjects in a given voxel.  This is the data we will eventually use for
our analysis.  To make sure we know what we are looking at, though, we want to see if we can reconstruct
event-related timecourses from this full timecourse data and make them look like those calculated by the
marsbar gui.  So far, we have been unsuccessful.  The event-related timecourses we calculate are very
different from those we obtain through the gui.  Can somebody please give advice about how to reconstruct
from full timecourse data the event-related timecourses (using percent signal change) that marsbar
outputs?  What might we be missing?

  Thank you in advance for your help!

-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/
Permalink | Reply | 0 comments |
headers
amyh | 29 Oct 04:09
Picon

Re: ROI data extraction error


Hi Matt,

To follow up with this question that was asked a few months ago, 

when I use the commands that you sent to Karen, I get:

D = mardo('SPM.mat');
names = image_names(D);
spm_vol(deblank(names(1,:)))

ans = 

    [1x1 struct]

I am getting the same error message that she got, and I was wondering if a
solution was found.

Thank you.
Amy

karen lythe wrote:
> 
> Hi, 
> I am having problems with MarsBaR when i try to 'extract ROI data'. I've
> seen on both the MarsBaR and SPM mailing lists that this is probably due
> to me moving files, so I have followed the steps from the FAQ regarding
> the error "can't open image file". Once i had specified where the files
> were i got the message that "all images in design appear to exist". I then
> went through the stages to extract the ROI data but get the same error.
(Continue reading)

Permalink | Reply | 0 comments |
headers
amyh | 29 Oct 07:17
Picon

Error in ==> maroi.getdata at 99


Hi, I'm getting an error when I try to extract my data.  The ROI looks okay,
and the design files are properly loaded.  But when I select "extract ROI
data (default)", it gives me the following message (below).  Any help would
be much appreciated.
Thank you,
Amy

Fetching data                           :                         1/1   ???
Cant open image file.

Error in ==> maroi.getdata at 99
  data = spm_sample_vol(data_imgs(i),...

Fetching data                           :                         1/1   ???
Cant open image file.

Error in ==> maroi.getdata at 99
  data = spm_sample_vol(data_imgs(i),...

Error in ==> maroi.get_marsy at 76
  [y vals vXYZ mat]  = getdata(o, VY);

Error in ==> marsbar at 949
marsY = get_marsy(o{:}, VY, sumfunc, 'v');

Error in ==> spm at 1276
evalin('base',CBs{v-1})

??? Error while evaluating uicontrol Callback.
(Continue reading)

Permalink | Reply | 2 comments |

Gmane