Dear exports,
I have a question about deactivation.
In my recent picture naming study, I found that so much deactivation existed in the images, even after group average by 3dttest.I had checked my processing procedure several times, there was no mistake. I had followed your group_analysis presentation.I did not know why it was, and felt much strange.
My study was block design,4 run and 7 minutes and 48 seconds for each.Each run had one condition,4 conditons totally.The stimulus was line-drawing picture,tasks naming.The baseline condition was passively seeing the nonsence picture.
The MRI was Siemens Magnetom Sonato Meastro Class,1.5T,TR=3000,FOV=220,slice number=20,slice thickness=6mm, base resolution=3.4*3.4*6.3mm.
My processing steps was followed:
2dImReg -input {$subj}_r{$run}+orig \
-basefile {$subj}_r1+orig -base 3 \
-prefix {$subj}_r{$run}_2dvr \
-dprefix {$subj}_r{$run}_2dvr_para.1D \
-rprefix {$subj}_r{$run}_2dvr_para_err.1D \
-debug
3dvolreg -verbose \
-base {$subj}_r1+orig'[3]' -clipit \
-tshift 2 -zpad 4 \
-1Dfile {$subj}_r{$run}_3dvr_para.1D \
-prefix {$subj}_r{$run}_vr \
{$subj}_r{$run}_2dvr+orig'[2..131]'
3dTsmooth -prefix {$subj}_r{$run}_vr_tsmooth \
-lin \
{$subj}_r{$run}_vr+orig'[0..129]'
3dAutomask -prefix mask_r{$run} {$subj}_r{$run}_blur+orig
3dcalc -a mask_r1+orig -b mask_r2+orig -c mask_r3+orig \
-d mask_r4+orig \
-expr 'step(a+b+c+d)' \
-prefix full_mask
3dTstat -prefix mean_r{$run} {$subj}_r{$run}_blur+orig
3dcalc -a {$subj}_r{$run}_blur+orig \
-b mean_r{$run}+orig \
-c full_mask+orig \
-expr "(a/b * 100) * c" \
-prefix scaled_r{$run}
3dDeconvolve -xout -input ./scaled_r1+orig \
-num_stimts 1 \
-stim_file 1 ../model/normal/{$subj}/{$subj}_r1hrf.1D -stim_label 1 one1 \
-glt 1 ../model/normal/{$subj}/{$subj}.1D -glt_label 1 one1 \
-censor ../model/normal/{$subj}/{$subj}_r1cencor.1D \
-fout -tout -polort 1 \
-progress 1000 \
-bucket ./{$subj}_r1func
.(run2)
.(run3)
.(run4)
So, What is the problem?
Formerly I had tried to 3dTcat four run and do analysis in your presentation way, but I could not sure if this way is suitable to my study of which each run had only one condition, and guessed this might be the cause. So next I do analysis each run separatly, but the difference was minor. And I thought that too much deactivation might be caused by sever head movement, I modeled the motion parameter in the GLM as baseline next,but the data became strange in another way,though deactivation had become less.Other attempts included add delay, but did benefit no more than before.
Can you experts tell me why it is? And if there was no problems, what does the deactivation means?
Any suggestions are appreciated.Thank you firstly.
your sincerely
Chunming Lu