Dear experts,
during the interpretation of my dual-regression results I got some weird activation in the white matter (so not for every component, but at least for one of them). I know that to some extend this is quite normal, but in this case it comprises hundreds of voxels! The original melodic ica component looks quite okay (in general the networks are absolutely plausible (DMN, frontoparietal, etc. ...) with some exceptions, but the dual regression output looks as described above (large extend white matter activation). (See the attachment for output of dual_regression (intensity already set to significance 0.99 and 1 in the case of my data!)).
Below I will describe my settings and inputfiles:
(A) melodic -i /data/listof_filtered_func_data.txt -o /data/outputfolder.gica --nobet(already done during preprocessing) --bgthreshold=10 -d 30 -a concat --mmthresh=0.5 --report --tr=2.2 --Oall
(B) afterwards I ran the dual regression:
dual_regression /data/outputfolder.gica/melodic_IC 1 -1 0 /data/OutputDualregression `cat /data/listof_filtered_func_data.txt`
(of course at this stage it is also possible to run the dual_regression all at once (I also tried this).
(C) finally I ran randomise:
randomise -i /data/OutputDualregression/dr_stage2_ic000x -o /data/OutputRandomise -n 500 -T -N -m /data/OutputDualregression/mask
In the designmatrix brainvolume and motionparameter were set as covariates (so the possible influence is corrected for).
Is there something strange in this procedere or am I using the wrong inputs?
What could possibly be the reason for large activations in the white matter?
Thank you for your (allways very helpful) answers!
Kind regards,
Stefan Lenzen
|