The problem is that all of your class=0 subjects have a 0 duration meaning that you cannot compute a duration slope for this class. In this special case, you can run a different-offset, same-slope analysis with --fsgd dur.fsgd doss Change your contrast matrix to 1 -1 0
On 11/14/19 9:47 AM, Ferraro, Pilar wrote: > > External Email - Use Caution > > Hi, > > I'm trying to run a surface based group analysis comparing patients > and HC, correcting for disease duration. > I enclose here the fsgd file and the design matrix. > I've followed all the steps of the tutorial > (http://surfer.nmr.mgh.harvard.edu/fswiki/FsTutorial/GroupAnalysisV6.0) with > no errors. > However, when I try to run the glm analysis with the following command: > > mri_glmfit.bin --y lh.dur.thickness.10.mgh --fsgd dur.fsgd dods --C > lh-Avg-thickness-dur-Cor.mtx --surf fsaverage lh --cortex --glmdir > lh.dur.glmdir > > I get the error: > > ERROR: matrix is ill-conditioned or badly scaled, condno = 1e+08 > -------------------------------- > Possible problem with experimental design: > Check for duplicate entries and/or lack of range of > continuous variables within a class. > > Of course there is no range for the continuous variable (disease > duration) in the control group, so I'm wondering how to overcome > this limit and proceed correctly with the glm analysis. > > Many thanks in advance for any help you will provide, > > Pilar > > > > > > > ------------------------------------------------------------------------ > *From:* Ferraro, Pilar > *Sent:* Monday, November 11, 2019 7:07 AM > *To:* [email protected] <[email protected]> > *Subject:* surface group analysis with qdec > Hi! > > I've previously run a comparison between patients and HC using qdec, > but now I'd like to correct the analysis for disease duration. > I've therefore generated a second qdec.table.dat file inserting the > disease duration column with the following structure: > fsid group duration > Input subj1 1 13 > Input subj2 0 > > (For healthy controls group (group 0) I've tried both options, leaving > the duration column blank or inserting 0 as a value). > I've also left the group.level file the same of the previous analysis. > However, when i open the new qdec.table.dat file with qdec I get > problems in correctly uploading the data. > In particular, I get the following error: > > ERROR: matrix is ill-conditioned or badly scaled, condno = 1e+08 > Possible problem with experimental design: > Check for duplicate entries and/or lack of range of continuous > variables within a class. > If you seek help with this problem, make sure to send: > This is the command line: > mri_glmfit.bin --y /home/ospite/Pilar/freesurfer/qdec/Untitled/y.mgh > --fsgd /home/ospite/Pilar/freesurfer/qdec/Untitled/qdec.fsgd dods > --glmdir /home/ospite/Pilar/freesurfer/qdec/Untitled --surf fsaverage > lh --label > /home/ospite/Pilar/freesurfer/fsaverage/label/lh.aparc.label --C > /home/ospite/Pilar/freesurfer/qdec/Untitled/contrasts/lh-Avg-Intercept-thickness.mat > > --C > /home/ospite/Pilar/freesurfer/qdec/Untitled/contrasts/lh-Diff-1-0-Intercept-thickness.mat > > Of course there is no range for the continuous variable (disease > duration) in the control group, so I'm wondering how to overcome this > and generate a correct qdec.table.dat file. > > Thanks in advance for any help! > > Pilar > > _______________________________________________ > Freesurfer mailing list > [email protected] > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer _______________________________________________ Freesurfer mailing list [email protected] https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
