Issue with use afni_proc.py

[danlittledan]

Hi,
I am running through an analysis with afni_proc.py and getting a problem with one subject that I run. The afni_proc.py command is the following:

afni_proc.py -subj_id sub10_ffa
-dsets FFArun*+orig.HEAD
-blocks despike tshift align tlrc volreg blur mask scale regress
-copy_anat anat+orig
-align_opts_aea -cost lpc+ZZ
-tlrc_base TT_N27+tlrc
-volreg_align_to MIN_OUTLIER
-volreg_align_e2a
-volreg_tlrc_warp
-blur_size 4.0
-regress_local_times
-regress_stim_times sub10_overstim/FFArun*.txt
-regress_stim_labels Face House
-regress_basis ‘BLOCK(16,1)’
-regress_opts_3dD
-gltsym ‘SYM: Face -House’
-glt_label 1 F-H
-regress_est_blur_errts
-execute

And the text output is:

subject ID : sub10_ffa
AFNI version : AFNI_18.3.15
AFNI package : linux_ubuntu_16_64
TRs removed (per run) : 0
num stim classes provided : 2
final anatomy dset : anat_final.sub10_ffa+tlrc.HEAD
final stats dset : stats.sub10_ffa+tlrc.HEAD
final voxel resolution : 3.000000 3.000000 3.000000

motion limit : 0.3
num TRs above mot limit : 0
average motion (per TR) : 0.0362701
max motion displacement : 2.59561
outlier limit : 0.1
average outlier frac (TR) : 0.000903081
num TRs above out limit : 0

num runs found : 2
num TRs per run : 172 172
TRs total : 344
degrees of freedom used : 16
degrees of freedom left : 328

num regs of interest : 2
num TRs per stim : 75 75
ave mot per sresp : 0.044615 0.040188

TSNR average : 167.225
global correlation (GCOR) : 0.0387909
anat/EPI mask Dice coef : 0.920801
maximum F-stat (masked) : 34.8551
blur estimates (ACF) : 0.726891 3.24651 11.4694
blur estimates (FWHM) : 0 0 0

I checked the QC file generated by proc.py, the result is the attached image 1, it’s abnormal compared with other subjects. It was normal to use the same script to analyze the data of other subjects.I have checked everything I can think of and cannot seem to find where this error is coming from.

I also added “-regress_censor_motion 0.2, -regress_censor_outliers 0.1, -regress_motion_per_run, -regress_apply_mot_types demean deriv” to the script.This subject’s data result did not do well either,the result is the attached image 2.

Thank you very much for any assistance,

Dan

[colmconn]

Have you thoroughly inspected the alignment of the EPI to the T1 images? Also, have you checked the brain masking?

[danlittledan]

I checked the alignment of the EPI and T1 images and the brain masking, the results are in attach files. It’s no problem.
And in this sample data, I did not get the ROI percent signal change I expected. It’s very confusing.

[rickr]

Hi Dan,

To be sure, what is your concern, the bright regions outside the brain? If so, those look like the eyes. Is there a reason to have stimulus correlated eye movement?

And it seems like there might be results in the visual cortex, which would make sense.

First check the same result, but with anat_w_skull_warped+tlrc as the underlay. That will make it more clear whether it is indeed the eyes.

• rick

[danlittledan]

Hi rick ,

First, I did a face fusiform gyrus localization experiment,using faces minus houses, we got a fusiform area that was sensitive to faces, but I don’t know why can’t get the percent signal change of FFA in the subject as my expect.
I’ve checked the result and it may have nothing to do with the visual cortex.(Please see the image attach file)
And I also provided a QC file result of data of my other normal subjects in attach file.

Thanks,
-Dan

[rickr]

Is it possible your subject fell asleep, or just did not really bother with the task? Do subjects respond to events with button presses?

Are you sure there is no mistake with the stimulus timing files?

Use Instacorr in the afni GUI, placing a seed in the fusiform gyrus. Does that show an expected activation pattern?

• rick

[danlittledan]

Hi rick,

Thanks for your help, I will check the timing file and other possibilities you mentioned.

Dan