 |
 |
Dear Christian, Thank you for your reply and to remind me to correct for comparisons! I managed to do the splitting, merging and randomizing of the data successfully. Now, I am wondering actually about the interpretation: If I ran fslmaths <pre> -sub <post> <output>, did FSL construct images pre-post or post-pre? And if I want to extract the values out ROIs with fslmeants from the DR2 merged files, what do these z-values reflect? Is this connectivity during rest of this ROI with the rest of the network (e.g. DMN)? Or should I see this as activation within this region? Furthermore, I also want to correct for grey matter atrophy, so I created a voxelwise EV via feat_gm_prepare. Do I have to demean this? And yes, how should I do this, since this is a 4D file? Thank you very much!! Best wishes Heidi ================================================ Dr. Heidi Jacobs Postdoc researcher Faculty of Health, Medicine and Life Sciences School for Mental Health and Neurosciences Division Cognitive Neuropsychiatry and Clinical Neurosciences Alzheimer Center Limburg [log in to unmask] www.maastrichtuniversity.nl www.heidijacobs.nl Dr. Tanslaan 12, 6229 ET Maastricht P.O. Box 616, 6200 MD Maastricht, The Netherlands T +31 43 38 84 090 F +31 43 38 84 092 ================================================ ________________________________________ Van: FSL - FMRIB's Software Library [[log in to unmask]] namens Christian F. Beckmann [[log in to unmask]] Verzonden: woensdag 3 oktober 2012 22:02 Aan: [log in to unmask] Onderwerp: Re: [FSL] Repost: question regarding pre and post comparison Hi Yes, use fslsplit or use fslmerge -t correctly on the individual maps. If you subtract the dr stage 2 maps pre vs post prior to running randomise you have an easier time getting randomise to do the right thing wrt permuting the data... Also, be aware that after splitting and running separate randomise runs on the data you will have to think about Bonferroni correcting the output to account for the multiple components you're investigating. The TFCE results will be corrected across voxels but the final results will need to be corrected across comparisons, too. hth Christian On 28 Sep 2012, at 17:08, Jacobs H (NP) <[log in to unmask]> wrote: > Hi, > > Thank you very much. > Yes, that is what I thought. I probably should do this with fslsplit, right? > I think it is also best to do this on the stage 2 images, before the subtraction, correct? > > Best > Heidi > > ================================================ > Dr. Heidi Jacobs > Postdoc researcher > Faculty of Health, Medicine and Life Sciences > School for Mental Health and Neurosciences > Division Cognitive Neuropsychiatry and Clinical Neurosciences > Alzheimer Center Limburg > [log in to unmask] > www.maastrichtuniversity.nl > www.heidijacobs.nl > > Dr. Tanslaan 12, 6229 ET Maastricht > P.O. Box 616, 6200 MD Maastricht, The Netherlands > T +31 43 38 84 090 F +31 43 38 84 092 > ================================================ > ________________________________________ > Van: FSL - FMRIB's Software Library [[log in to unmask]] namens Jeanette Mumford [[log in to unmask]] > Verzonden: vrijdag 28 september 2012 16:14 > Aan: [log in to unmask] > Onderwerp: Re: [FSL] Repost: question regarding pre and post comparison > > Hi, > > Yes, you must run separate 2-sample t-tests for each component. So, each 2-sample t-test will have an input 4D file with 40 images (1 paired difference within-subject within a specific component). > > Cheers, > Jeanette > > On Fri, Sep 28, 2012 at 7:02 AM, Jacobs H (NP) <[log in to unmask]<mailto:[log in to unmask]>> wrote: > Hi, > > I will try to explain: > > I first did the ICA on all the data (2 groups of 20 persons and 2 time points, so 80 data sets). > Then I did the dual regression on the filtered func files with the beckmann template as a mask (this contains 8 resting state components + CSF + white matter). > So, in stage 2 of the dual regression, I have : > - files for each subject (each subject files has the 10 components in it): so 80 files > - files for each component (containing all the subjects in it): so 10 files > > In order to compare the difference in 2 time points, I subtracted the individual stage 2 files: resulted in 40 files, which each containing the time course of the 10 components for the difference of the two conditions. > I than merged the 40 files in one 4D files, but now this has 400 time points (40 people x 10 timecourses (one for each component)). > > I don't know if there is a way to split the files in different ones for each component?? Or whether it is safe to do this? > > I now would like to go to stage 3 dual regression with these subtracted files, so that I can get the statistics (t-stat). Usually this results in a file for each contrast for each component. > > I hope this helps and that somebody could help me :) > > Thanks! > Heidi > > > > ================================================ > Dr. Heidi Jacobs > Postdoc researcher > Faculty of Health, Medicine and Life Sciences > School for Mental Health and Neurosciences > Division Cognitive Neuropsychiatry and Clinical Neurosciences > Alzheimer Center Limburg > [log in to unmask]<mailto:[log in to unmask]> > www.maastrichtuniversity.nl<http://www.maastrichtuniversity.nl> > www.heidijacobs.nl<http://www.heidijacobs.nl> > > Dr. Tanslaan 12, 6229 ET Maastricht > P.O. Box 616, 6200 MD Maastricht, The Netherlands > T +31 43 38 84 090<tel:%2B31%2043%2038%2084%20090> F +31 43 38 84 092<tel:%2B31%2043%2038%2084%20092> > ================================================ > ________________________________________ > Van: FSL - FMRIB's Software Library [[log in to unmask]<mailto:[log in to unmask]>] namens Jeanette Mumford [[log in to unmask]<mailto:[log in to unmask]>] > Verzonden: vrijdag 28 september 2012 13:49 > Aan: [log in to unmask]<mailto:[log in to unmask]> > Onderwerp: Re: [FSL] Repost: question regarding pre and post comparison > > Hi, > > I'm afraid I don't understand the problem. If you're analyzing 10 components then you will need to run 10 separate 2 sample t-tests. > > I'm also not sure what the Beckmann template is. > > Cheers, > Jeanette > > On Fri, Sep 28, 2012 at 5:35 AM, Jacobs H (NP) <[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>>> wrote: > Dear Jeanette, > > Thank you very much for your help. > However, I came across another error: the number of time points in the merged 4D file does not match my design matrix. > Probably this is because I used the Beckmann template in the dual regression (providing the components). > So, now I have 40 people x 10 components, which is different from my design matrix (having 2 groups of 20 people). > > How should I proceed to get separate files for each component? And still be able to proceed to stage 3 from DR do the 2 sample t-test on the subtracted images (condition-condition2)? > > Thank you very much! > > Best > Heidi > > ================================================ > Dr. Heidi Jacobs > Postdoc researcher > Faculty of Health, Medicine and Life Sciences > School for Mental Health and Neurosciences > Division Cognitive Neuropsychiatry and Clinical Neurosciences > Alzheimer Center Limburg > [log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>> > www.maastrichtuniversity.nl<http://www.maastrichtuniversity.nl><http://www.maastrichtuniversity.nl> > www.heidijacobs.nl<http://www.heidijacobs.nl><http://www.heidijacobs.nl> > > Dr. Tanslaan 12, 6229 ET Maastricht > P.O. Box 616, 6200 MD Maastricht, The Netherlands > T +31 43 38 84 090<tel:%2B31%2043%2038%2084%20090><tel:%2B31%2043%2038%2084%20090> F +31 43 38 84 092<tel:%2B31%2043%2038%2084%20092> > ================================================ > ________________________________________ > Van: FSL - FMRIB's Software Library [[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>>] namens Jeanette Mumford [[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>>] > Verzonden: vrijdag 28 september 2012 1:11 > Aan: [log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>> > Onderwerp: Re: [FSL] Repost: question regarding pre and post comparison > > Hi, > > you need to first create a 4D image and use that as input into randomise fslmerge -t can be used for this purpose. See the the help for complete usage. > > Cheers, > Jeanette > > On Thu, Sep 27, 2012 at 3:36 PM, Jacobs H (NP) <[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>><mailto:[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>>>> wrote: > Hi, > > I managed to subtract all the pre and post files of all my subjects from stage 2 from DR (where I used the Beckmann template). Furthermore, I also made a .mat and .con file for an 2 group unpaired t-test of these files. > I now would like to run the statistics on these ' subtracted' images and then find out in which components there is a group difference for the condition1-2. > Should I use randomise for this? I tried it, but it won't work: > > randomise -i A01_subtracted_DR2.nii.gz A02_subtracted_DR2.nii.gz A03_subtracted_DR2.nii.gz A04_subtracted_DR2.nii.gz A05_subtracted_DR2.nii.gz A06_subtracted_DR2.nii.gz A07_subtracted_DR2.nii.gz A08_subtracted_DR2.nii.gz A09_subtracted_DR2.nii.gz A10_subtracted_DR2.nii.gz A11_subtracted_DR2.nii.gz A12_subtracted_DR2.nii.gz A13_subtracted_DR2.nii.gz A14_subtracted_DR2.nii.gz A15_subtracted_DR2.nii.gz A16_subtracted_DR2.nii.gz A17_subtracted_DR2.nii.gz A18_subtracted_DR2.nii.gz A19_subtracted_DR2.nii.gz A20_subtracted_DR2.nii.gz J01_subtracted_DR2.nii.gz J02_subtracted_DR2.nii.gz J03_subtracted_DR2.nii.gz J04_subtracted_DR2.nii.gz J06_subtracted_DR2.nii.gz J07_subtracted_DR2.nii.gz J08_subtracted_DR2.nii.gz J09_subtracted_DR2.nii.gz J10_subtracted_DR2.nii.gz J11_subtracted_DR2.nii.gz J12_subtracted_DR2.nii.gz J13_subtracted_DR2.nii.gz J14_subtracted_DR2.nii.gz J15_subtracted_DR2.nii.gz J16_subtracted_DR2.nii.gz J17_subtracted_DR2.nii.gz J18_subtracted_DR2.nii.gz J19_subtracted_DR2.nii.gz J20_subtracted_DR2.nii.gz J21_subtracted_DR2.nii.gz -o Output_DR3_subtraction -d model_subtractedDR2.mat -t model_subtractedDR2.con -n 5000 > > Could you tell me what I am doing wrong and how I should proceed on this? > Thank you very much! > > Heidi > > > ================================================ > Dr. Heidi Jacobs > Postdoc researcher > Faculty of Health, Medicine and Life Sciences > School for Mental Health and Neurosciences > Division Cognitive Neuropsychiatry and Clinical Neurosciences > Alzheimer Center Limburg > [log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>><mailto:[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>>> > www.maastrichtuniversity.nl<http://www.maastrichtuniversity.nl><http://www.maastrichtuniversity.nl><http://www.maastrichtuniversity.nl> > www.heidijacobs.nl<http://www.heidijacobs.nl><http://www.heidijacobs.nl><http://www.heidijacobs.nl> > > Dr. Tanslaan 12, 6229 ET Maastricht > P.O. Box 616, 6200 MD Maastricht, The Netherlands > T +31 43 38 84 090<tel:%2B31%2043%2038%2084%20090><tel:%2B31%2043%2038%2084%20090><tel:%2B31%2043%2038%2084%20090> F +31 43 38 84 092<tel:%2B31%2043%2038%2084%20092> > ================================================ > ________________________________________ > Van: FSL - FMRIB's Software Library [[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>><mailto:[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>>>] namens Stephen Smith [[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>><mailto:[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>>>] > Verzonden: donderdag 27 september 2012 10:22 > Aan: [log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>><mailto:[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>>> > Onderwerp: Re: [FSL] Repost: question regarding pre and post comparison > > HI - yes it's the stage 2 outputs you want. > Cheers. > > > On 27 Sep 2012, at 09:18, Jacobs H (NP) wrote: > > Hi, > > I got fslmaths working! > But should I use the dr_stage2_subject[#SUB].nii.gz from stage 2 dual regression? Thus: subtract the two measurements for each subject and feed this stage 3 DR (with the subtracted maps) for doing a 2 sample t-test? > > Thanks > Heidi > ================================================ > Dr. Heidi Jacobs > Postdoc researcher > Faculty of Health, Medicine and Life Sciences > School for Mental Health and Neurosciences > Division Cognitive Neuropsychiatry and Clinical Neurosciences > Alzheimer Center Limburg > [log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>><mailto:[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>>><mailto:[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>><mailto:[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>>>> > www.maastrichtuniversity.nl<http://www.maastrichtuniversity.nl><http://www.maastrichtuniversity.nl><http://www.maastrichtuniversity.nl> > www.heidijacobs.nl<http://www.heidijacobs.nl><http://www.heidijacobs.nl><http://www.heidijacobs.nl> > > Dr. Tanslaan 12, 6229 ET Maastricht > P.O. Box 616, 6200 MD Maastricht, The Netherlands > T +31 43 38 84 090<tel:%2B31%2043%2038%2084%20090><tel:%2B31%2043%2038%2084%20090><tel:%2B31%2043%2038%2084%20090> F +31 43 38 84 092<tel:%2B31%2043%2038%2084%20092> > ================================================ > ________________________________________ > Van: FSL - FMRIB's Software Library [[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>><mailto:[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>>>] namens Jeanette Mumford [[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>><mailto:[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>>>] > Verzonden: maandag 24 september 2012 16:35 > Aan: [log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>><mailto:[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>>> > Onderwerp: Re: [FSL] Repost: question regarding pre and post comparison > > Hi, > > Yes, to test how the pre/post differences compare between your two groups, you would first subtract the pre/post within subject and then enter these differences into a 2-sample t-test. > > Other within pre or within post difference between your groups would be run with separate two-sample t-tests. > > Cheers, > Jeanette > > On Mon, Sep 24, 2012 at 3:35 AM, Jacobs H (NP) <[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>><mailto:[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>>><mailto:[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>><mailto:[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>>>>> wrote: > Dear FSL team, > > I am struggling a bit with the statistics for my resting state analyses. > I have two groups (each 20 subejcts) and a pre and post measurement. I already ran the ICA and now would like to do the dual regression. > In this regression I would like to investigate the main effect of group, condition (pre/post) and the interaction (group differences on the difference between post and pre). > > Should I do a 2by2 anova with 4 EV's (group1 timepoint 1; group 1 timepoint 2; group 2 timepoint 1; group 2 timepoint 2). > > And then create the following contrast file > > main effect group: 1 1 -1 -1 > main effect time 1 -1 1 -1 > interaction: 1 -1 -1 1 > > However, I read somewhere that I should subtract both conditions for each individual and then compare the two group on this 'subtracted image'?? > Is this correct? > > Any advice would be greatly appreciated!! > > Thanks! > > Best > Heidi > > > > > > > ================================================ > Dr. Heidi Jacobs > Postdoc researcher > Faculty of Health, Medicine and Life Sciences > School for Mental Health and Neurosciences > Division Cognitive Neuropsychiatry and Clinical Neurosciences > Alzheimer Center Limburg > [log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>><mailto:[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>>><mailto:[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>><mailto:[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>>>> > www.maastrichtuniversity.nl<http://www.maastrichtuniversity.nl><http://www.maastrichtuniversity.nl><http://www.maastrichtuniversity.nl><http://www.maastrichtuniversity.nl> > www.heidijacobs.nl<http://www.heidijacobs.nl><http://www.heidijacobs.nl><http://www.heidijacobs.nl><http://www.heidijacobs.nl> > > Dr. Tanslaan 12, 6229 ET Maastricht > P.O. Box 616, 6200 MD Maastricht, The Netherlands > T +31 43 38 84 090<tel:%2B31%2043%2038%2084%20090><tel:%2B31%2043%2038%2084%20090><tel:%2B31%2043%2038%2084%20090><tel:%2B31%2043%2038%2084%20090> F +31 43 38 84 092<tel:%2B31%2043%2038%2084%20092> > ================================================ > > > > --------------------------------------------------------------------------- > 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<tel:%2B44%20%280%29%201865%20222726><tel:%2B44%20%280%29%201865%20222726> (fax 222717) > [log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>><mailto:[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>>><mailto:[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>><mailto:[log in to unmask]<mailto:[log in to unmask]><mailto:[log in to unmask]<mailto:[log in to unmask]>>>> http://www.fmrib.ox.ac.uk/~steve > ---------------------------------------------------------------------------