 |
 |
Dear FSL Experts,
We are combining RS-fMRI data from three different studies to explore the relation between a behavioural covariate to RS-activity following administration of different psychoactive drugs. Because these different studies have a different number of scans per subject (5-8 scans per condition, 2-3 conditions per subject), we are performing a dual regression (including the three behavioural covariates) for each individual subject using the RSN templates (Smith et al.) and will feed the resulting z-stats into a group-level analysis with randomise.
For 40 of the 48 subjects, the dual regression worked and completed within 24 hours (some as quick as two hours). For the remaining 8 subjects, the dual regression is still running (more than three weeks now). Within the Activity Monitor I can see there is activity (processes with 'randomise' have ~100% CPU usage), but there is no additional output generated. I have decided to restart the analyses once already after two weeks, but the same thing happens. The command-line call and processing of the files is exactly the same as for the other 40 subjects. The 8 subjects appear to be a random subset of the 48, although for some there appears to be some more variation on the behavioural measure. Other than that the data looks normal. We also reran the analysis with only 2 permutations, which also was not finished after a day.
We think the analysis may have entered a loop or something. Do you know if this can still be normal, or is something wrong (either with the data or my approach)? What can I do to run diagnostics? I can upload the data if that's helpful.
Thanks in advance for your help!
Daniël Kleinloog
PhD Candidate
Centre for Human Drug Research / Leiden University Medical Centre
Leiden, the Netherlands
--
Command:
dual_regression rsn20.nii.gz 1 Design.mat Design.con 500 dr_rsn20 Input1 Input2 ... InputN
Design.mat:
/NumWaves 7
/NumPoints 14
/PPheights 1.000000e+00 1.000000e+00 1.000000e+00 1.200000e+01 3.000000e+00 6.000000e+00 2.000000e+00
/Matrix
1.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00
1.000000e+00 0.000000e+00 0.000000e+00 -6.000000e+00 0.000000e+00 2.000000e+00 0.000000e+00
1.000000e+00 0.000000e+00 0.000000e+00 2.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00
1.000000e+00 -1.000000e+00 0.000000e+00 1.000000e+00 0.000000e+00 -1.000000e+00 0.000000e+00
1.000000e+00 -1.000000e+00 0.000000e+00 -3.000000e+00 0.000000e+00 -1.000000e+00 0.000000e+00
1.000000e+00 0.000000e+00 0.000000e+00 6.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00
1.000000e+00 -1.000000e+00 0.000000e+00 -1.000000e+00 0.000000e+00 -4.000000e+00 0.000000e+00
1.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00
1.000000e+00 0.000000e+00 1.000000e+00 0.000000e+00 -1.000000e+00 0.000000e+00 2.000000e+00
1.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 -1.000000e+00 0.000000e+00 0.000000e+00
1.000000e+00 0.000000e+00 1.000000e+00 0.000000e+00 1.000000e+00 0.000000e+00 2.000000e+00
1.000000e+00 0.000000e+00 1.000000e+00 0.000000e+00 1.000000e+00 0.000000e+00 2.000000e+00
1.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 2.000000e+00 0.000000e+00 0.000000e+00
1.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 1.000000e+00 0.000000e+00 0.000000e+00
Design.con:
/ContrastName1 Behavioural measure 1 (active - placebo)
/ContrastName2 Behavioural measure 2 (active - placebo)
/ContrastName3 Behavioural measure 3 (active - placebo)
/NumWaves 7
/NumContrasts 3
/PPheights 8.112320e-01 2.831488e+00 2.162832e+00
/RequiredEffect 13.596 14.751 12.970
/Matrix
0.000000e+00 1.000000e+00 -1.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00
0.000000e+00 0.000000e+00 0.000000e+00 1.000000e+00 -1.000000e+00 0.000000e+00 0.000000e+00
0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 1.000000e+00 -1.000000e+00
System:
FSL 5..0.4 on a MacBook Pro with Mac OS 10.8.4 with a 2.3 GHz intel core i7 processor and 16Gb of memory.