MarsBaR ROI toolbox mailing list ()

raw_data: extract with or without a design?

Hello,

I'm using MarsBar to compare two groups in a specific ROI. However, when extracting raw_data from the two-sampled design (export data) I get different results as compared to the raw_data from individual image extractions from the same con-files. In this last case I'm extracting the raw_data without a design and thus without any scaling (extract ROI, full options, no design).

I can understand that this will lead to differences in raw_data wihtin an individual, but I don't understand why the distribution between the indivduals are very different.

In addition, when comparing 'means of the two groups of the raw_data extracted from the two-sampled design' with 'the means of the raw_data extraction without a design' gave different distributions of the groups and different results. How is this possible? And, in order to test for a correlation with behavior data, which raw_data will I use ...the two-sampled or the 'real' raw data?

Hope you understand the problem,

Kind regards,

Martijn

Martijn Mulder, PhD Student

Rudolf Magnus Institute of Neuroscience
University Medical Center Utrecht | Child & Adolescent Psychiatry

Neuroimaging department

E: m.mulder-7 <at> umcutrecht.nl

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV

_______________________________________________
Marsbar-users mailing list
Marsbar-users <at> lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/marsbar-users

headers

golcud@mail.rockefeller.edu | 4 Dec 15:55

mismatch of analysis with tutorial

Hi, 

I am a new user of marsbar. I was going through the tutorial, and
the results of my analysis do not match the results given in the
tutorial. Specificly, for 

sess1 contrast value  2.27 t statistic 4.85 p values 0.000002
sess3 contrast value  3.19 t statistic 4.49 p values 0.000008

compared to the tutotorial values of 

sess1 contrast value  2.40 t statistic 5.36 p values 0.000000
sess3 contrast value  3.08 t statistic 4.45 p values 0.000010

Since I have SPM5, I had to do the preprocessing by hand, but as
far as I can tell, there is no difference between my analysis and
the analysis in the tutorial. 

Any idea where this difference might be coming from?

Thanks,

Doruk 

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV

headers

Matthew Brett | 4 Dec 18:06

Re: mismatch of analysis with tutorial

Hi,

> I am a new user of marsbar. I was going through the tutorial, and
> the results of my analysis do not match the results given in the
> tutorial. Specificly, for
>
> sess1 contrast value  2.27 t statistic 4.85 p values 0.000002
> sess3 contrast value  3.19 t statistic 4.49 p values 0.000008
>
> compared to the tutotorial values of
>
> sess1 contrast value  2.40 t statistic 5.36 p values 0.000000
> sess3 contrast value  3.08 t statistic 4.45 p values 0.000010

Hmm - I am guessing that what is happening that SPM5 is not picking up
on the .mat files that give the orientation of the images L-R.  Could
you try flipping the smoothed images (using the SPM display, -1 zoom
in X, reorient feature) and do the analysis again?

Thanks a lot,

Matthew

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV

headers

Matthew Brett | 5 Dec 16:25

Re: raw_data: extract with or without a design?

Hi,

Sorry to be slow to reply.

> I'm using MarsBar to compare two groups in a specific ROI. However, when
> extracting raw_data from the two-sampled design (export data) I get
> different results as compared to the raw_data from individual image
> extractions from the same con-files.

Did you see the earlier thread on something similar?

http://permalink.gmane.org/gmane.comp.graphics.spm.marsbar/397

Best,

Matthew

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV

headers

golcud@mail.rockefeller.edu | 5 Dec 18:11

Re: mismatch of analysis with tutorial

I do not think this is the cause since as soon as I did this
flip, the displays became flipped compared to the figures printed
in the tutorial (such as Fig. 8 or Fig. 10 of the tutorial), so I
am assuming the original orientation, which matched these
examples more closely, was correct.

---------Included Message----------
>Date: 4-Dec-2006 12:06:58 -0500
>From: "Matthew Brett" <matthew.brett <at> gmail.com>
>To: "golcud <at> mail.rockefeller.edu" <golcud <at> mail.rockefeller.edu>
>Cc: <marsbar-users <at> lists.sourceforge.net>
>Subject: Re: [Marsbar-users] mismatch of analysis with tutorial
>
>Hi,
>
>> I am a new user of marsbar. I was going through the tutorial, and
>> the results of my analysis do not match the results given in the
>> tutorial. Specificly, for
>>
>> sess1 contrast value  2.27 t statistic 4.85 p values 0.000002
>> sess3 contrast value  3.19 t statistic 4.49 p values 0.000008
>>
>> compared to the tutotorial values of
>>
>> sess1 contrast value  2.40 t statistic 5.36 p values 0.000000
>> sess3 contrast value  3.08 t statistic 4.45 p values 0.000010
>
>Hmm - I am guessing that what is happening that SPM5 is not
picking up
>on the .mat files that give the orientation of the images L-R. 
Could
>you try flipping the smoothed images (using the SPM display, -1 zoom
>in X, reorient feature) and do the analysis again?
>
>Thanks a lot,
>
>Matthew
>
>-------------------------------------------------------------------------
>Take Surveys. Earn Cash. Influence the Future of IT
>Join SourceForge.net's Techsay panel and you'll get the chance
to share your
>opinions on IT & business topics through brief surveys - and
earn cash
>http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
>_______________________________________________
>Marsbar-users mailing list
>Marsbar-users <at> lists.sourceforge.net
>https://lists.sourceforge.net/lists/listinfo/marsbar-users
>
>
---------End of Included Message----------

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV

headers

Susan Meißner | 20 Dec 13:40

Putamen ROI

Dear Matthew, dear marsbar users,

We have run two experiments (TR =1.5, 18 Slices) and are interested in comparing the activities in the right and left putamen between these two experiments. We realigned, normailized and smoothed our data with SPM99. Furthermore the design was specified and contrast of interest, that reflects the individual putamen activity, was calculated for each subject (First Level Analysis). For comparing putamen activity between the two expriments we want to run a roi analysis in left and right putamen. Based on the tutorial first we specified the design (basis models à two sample t-test). Then the data for left or right putamen were extracted. We used the anatomical ROIs from the AAL Library. The model was estimated (Estimate Results) and the contrast (Experiment 1 > Experiment 2) was calculated.

We are new to marsbar. Therefore we are unsure if this procedure is correct? In the tutorial you wrote that the putamen does not vary much between subjects after spatial normalisation.

Are the anatomical ROIs from the AAL Library also a good approximation to our data? Or would it be better to define the ROIs using structural scans for each participant?

We have another problem with the example data of the tutorial. We use version marsbar-0.38.2 and can not find the preprocessing script.

Thank you

Der WEB.DE SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen!
http://smartsurfer.web.de/?mc=100071&distributionid=000000000066

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV

_______________________________________________
Marsbar-users mailing list
Marsbar-users <at> lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/marsbar-users

headers

Martijn Mulder | 23 Dec 21:33

compare ROI over 3 groups

Hi,

We have three groups and created ROI’s on basis of the control group.

What would you suggest for a ROI comparison between 3 groups?

We now using the two-sampled t-test design created with SPM. >From these designs we extracted the raw-data for the groups and put them in a one-way ANOVA with a statistical program and did a bonf. corrected post hoc test.

Are there other ways to do this?

I noticed that group means are dependend on how the SPM two-sampled design was specified:

For example: two-samped design for group1 (controls) & group2 (patients) gives different raw-data compare to a design for group1 (patients) & group2 (controls). So, the design depends on the first group in the design, which is strange, since we didn’t use contrasts in the design or in MarsBar.

How is this possible?

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV

_______________________________________________
Marsbar-users mailing list
Marsbar-users <at> lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/marsbar-users

Mon	Tue	Wed	Thu	Fri	Sat	Sun

				1	2	3
4	5	6	7	8	9	10
11	12	13	14	15	16	17
18	19	20	21	22	23	24
25	26	27	28	29	30	31

17	February 2012
8	January 2012
8	December 2011
29	November 2011
28	October 2011
19	September 2011
15	August 2011
11	July 2011
38	June 2011
39	May 2011
27	April 2011
20	March 2011
26	February 2011
33	January 2011
13	December 2010
16	November 2010
9	October 2010
22	September 2010
4	August 2010
14	July 2010
24	June 2010
10	May 2010
30	April 2010
17	March 2010
3	February 2010
8	January 2010
7	December 2009
4	November 2009
13	October 2009
12	September 2009
12	August 2009
27	July 2009
12	June 2009
7	May 2009
19	April 2009
14	March 2009
13	February 2009
21	January 2009
9	December 2008
17	November 2008
12	October 2008
6	September 2008
9	August 2008
9	July 2008
20	June 2008
11	May 2008
13	April 2008
7	March 2008
5	February 2008
3	January 2008
10	December 2007
19	November 2007
15	October 2007
17	September 2007
24	August 2007
29	July 2007
29	June 2007
21	May 2007
30	April 2007
30	March 2007
19	February 2007
13	January 2007
7	December 2006
5	November 2006
14	October 2006
18	September 2006
25	August 2006
24	July 2006
6	June 2006
18	May 2006
11	April 2006
9	March 2006
19	February 2006
31	January 2006
11	December 2005
19	November 2005
12	October 2005
14	September 2005
30	August 2005
11	July 2005
23	June 2005
13	May 2005
16	April 2005
23	March 2005
38	February 2005
17	January 2005
1	January 2003