Hi Anderson,
Thanks a lot for your help.
About the DVARS: The “as is” approach is clear, but the second approach is somewhat confusing. When should it be used?
Regards
Yacila
Von: FSL - FMRIB's Software Library [mailto:[log in to unmask]] Im Auftrag von Anderson M. Winkler
Gesendet: Montag, 19. September 2016 09:07
An: [log in to unmask]
Betreff: Re: [FSL] AW: [FSL] AW: [FSL] Rank deficient design within subjects PALM
Hi Yacila,
There isn't anything particularly odd in this command line. That error message is often caused by statistics that, for some reason, ended up being calculated as NaN or Inf, which generally is due to some issue with the design or contrasts (e.g., testing whether an IC is different than zero, when we know for sure already that they are different than zero).
About the repeated DVARS, yes, it can be entered "as is" or with the first visit multiplied by -1; the interpretation differs, though. Entering "as is" means treating DVARS differences as nuisance, whereas entering with the sign reversed means testing if the average across visits is a nuisance.
All the best,
Anderson
On 17 September 2016 at 18:21, Yacila Isabela Deza Araujo <[log in to unmask]> wrote:
Dear Anderson,
My command was:
palm -i dr_stage2_ic0000.nii. -i dr_stage2_ic0001.nii. -i dr_stage2_ic0002.nii -i dr_stage2_ic0003.nii -i dr_stage2_ic0004.nii -i dr_stage2_ic0005.nii -i dr_stage2_ic0006.nii -i dr_stage2_ic0007.nii. -i dr_stage2_ic0008.nii -i dr_stage2_ic0009.nii –m mask.nii -d within_all.mat -t within_all.con -eb EB.grp. -within -whole -corrcon -corrmod -nouncorrected -n 500 -approx tail -T -save1-p -o Palm_results
I was reading and my design could be rank deficient because I add gender. I am creating a new one without gender, but with the DVARS measures (since they are significantly different between visits). You mentioned that I need to multiple just the value of the DVARS from the first visit by -1 and the other two visits remain with the same sign, is this correct?
Thanks again
Yacila
Von: FSL - FMRIB's Software Library [mailto:[log in to unmask]] Im Auftrag von Anderson M. Winkler
Gesendet: Samstag, 17. September 2016 18:56
An: [log in to unmask]
Betreff: Re: [FSL] AW: [FSL] Rank deficient design within subjects PALM
Hi Yacila,
Could you show the full command line, design and contrasts for this particular run? Thanks.
All the best,
Anderson
On 17 September 2016 at 17:08, Yacila Isabela Deza Araujo <[log in to unmask]> wrote:
Hi Anderson,
Thanks a lot for your invaluable help. I found another problem. After running PALM, The output was not saved and I got the following error in Matlab:
Error using griddedInterpolant
The coordinates of the input points must be finite values; Inf and NaN are not permitted.
Error in interp1>Interp1D (line 335)
F = griddedInterpolant(X,V(:,1),method);
Error in interp1 (line 220)
Vq = Interp1D(X,V,Xq,method);
Error in palm_pareto>andersondarling (line 195)
i2 = interp1(i1,ptable,A2,'linear','extrap');
Error in palm_pareto (line 127)
A2pval = andersondarling(gpdpvals(ytail,apar,kpar),kpar);
Error in palm_saveall (line 245)
Ptosave =
palm_pareto(plm.G{y}{m}{c},plm.Gmax{y}{m}{c},false,opts.accel.tail_thr,opts.accel.G1out);
Error in palm_core (line 2476)
palm_saveall(plm,opts);
Error in palm (line 81)
palm_core(varargin{:});
What does it mean?. I need to check my data I think.
Thanks a lot for your time
Yacila
Von: FSL - FMRIB's Software Library [mailto:[log in to unmask]] Im Auftrag von Anderson M. Winkler
Gesendet: Samstag, 10. September 2016 10:20
An: [log in to unmask]
Betreff: Re: [FSL] Rank deficient design within subjects PALM
Hi Yacila,
The subject-specific EVs already code everything that doesn't change between the visits (such as sex). About the dvars: yes, can include them as nuisance.
In designs as this, it worth asking whether these nuisance variables would be affecting the individual measurements at each visit (in which case they go "as is", as long as there is no rank deficiency as in your case), or whether these would be affecting the differences between the visits (in which case they go with the signs alternating, i.e., multiply the 1st visit by -1, and leave the others unchanged).
All the best,
Anderson
On 9 September 2016 at 20:33, Yacila Isabela Deza Araujo <[log in to unmask]> wrote:
Hi Anderson,
I have a short question. I am running repeated measures with PALM (3 run per participants). My call is:
Palm –i input 1 –i input 2… -d design.mat –t design.con –eb –eb.grp –within –whole –corrcon –cormod –n 5000 –T –save1-p –o results
The design matrix has the form of:
Group EV1 EV2 EV3 …. EVn EVn+1
1 1 1 1 0 demean gender demean age
1 -1 0 1 0
1 0 -1 1 0
2 1 1 0 1
2 -1 0 0 1
2 0 -1 0 1
etc
As you indicated in previous messages.
I got this message before every permutation:
Matrix is close to singular or badly scaled. Results may be inaccurate. RCOND…
I read that it may be caused when two columns are identical. I checked my matrix (even for possible correlation between age and gender) and there are no identical columns. Any other suggestion?
The second issue is that my motion parameters are significant (more movement in the third session or more movement under one intervention) I used DVARS and FD. Should I include them also as nuisance regressors? If yes, all three of them (one per session) and demeaned?
Thanks a lot for your help and guidance
Yacila
