Dear Colleagues,

I have been trying to run the preprocessing script on my resting state data (v. 1.1b), and I keep running into errors during the nuisance signal regression stage. What I keep getting is below:

Modifying model file
Running feat model
Not enough data in /Users/Duke/scripts/test/01626306/func/nuisance/csf.1D
Running film to get residuals
Log directory is: /Users/Duke/scripts/test/01626306/func/nuisance/stats


An exception has been thrown
Unable to open /Users/Duke/scripts/test/01626306/func/nuisance/nuisance.mat
++ 3dTstat: AFNI version=AFNI_2009_12_31_1431 (Jul 23 2010) [64-bit]
++ Authored by: KR Hammett & RW Cox
** FATAL ERROR: Can't open dataset /Users/Duke/scripts/test/01626306/func/nuisance/stats/res4d.nii.gz
** Program compile date = Jul 23 2010
++ 3dcalc: AFNI version=AFNI_2009_12_31_1431 (Jul 23 2010) [64-bit]
++ Authored by: A cast of thousands
** FATAL ERROR: can't open dataset /Users/Duke/scripts/test/01626306/func/nuisance/stats/res4d.nii.gz
** Program compile date = Jul 23 2010
** ERROR (nifti_image_read): failed to find header file for '/Users/Duke/scripts/test/01626306/func/4DI001_res'
** ERROR: nifti_image_open(/Users/Duke/scripts/test/01626306/func/4DI001_res): bad header info
Error: failed to open file /Users/Duke/scripts/test/01626306/func/4DI001_res
ERROR: Could not open image /Users/Duke/scripts/test/01626306/func/4DI001_res
Image Exception : #22 :: Failed to read volume /Users/Duke/scripts/test/01626306/func/4DI001_res
terminate called after throwing an instance of 'RBD_COMMON::BaseException'
/Users/Duke/scripts/5_nuisance.sh: line 95: 63891 Abort trap flirt -ref ${reg_dir}/standard -in ${func_dir}/${rest}_res -out ${func_dir}/${rest}_res2standard -applyxfm -init ${reg_dir}/example_func2standard.mat -interp trilinear

I'm guessing it has something to do with nuisance.mat, and I don't seem to get that file after I run the script. If you have any ideas as to why I keep getting this error, and particularly any ways that I could fix this, I would be greatly appreciative.

Many thanks in advance and all the best.

Duke

Hi Duke,

It seems that the problem is in this line, which will make the rest of the nuisance removal spit out erros.

Not enough data in /Users/Duke/scripts/test/01626306/func/nuisance/csf.1D

You get this error when there is a problem with your csf mask. It does not include any voxels, making it unable to extract signal from it to remove. I can think of a couple of problems that might have caused this:

- The skullstripping of the anatomical scan removed the values in CSF as if they were skull, we've seen this happen a couple of times. You could resolve this by using bet (FSL) instead of 3dskullstrip. Play with the values abit to get the optimal result.
- The segmentation of the anatomical did not run properly, leaving you with a bad CSF image. Rerunning the segmentation might do it for you.

Hope this helps!
Maarten

Hi Duke,

I have met with similar problems before. I figured out that my problem was not enough memory on the computer. Approximately 6~8 G free would be better.
If you have more than 16G free memory. Then memory would not be a problem for data of about 200 time points.

If you don't have so much free memory, try using smaller data, such as 100 points by changing the parameters in the batch list setting
/your-directory/NewYork_a /your-directory/NewYork_a/NewYork_a_subjects.txt 0 99 100 2
like this.

If this works, then that is the problem.

Good luck.

B.T.W., many thanks to Maarten for remind me this since I initially replied by email but the content was not shown here.

Liang

Hi Ningqing,

your response is empty. Don't know if you meant if like that.

Best,
Maarten

Many thanks for the responses. I adjusted the segmentation parameters and that did the trick.

All the best,

Duke

Hi,

Just want to mention this potential bug is due to the default usage of '-normal_vol' option in 3dskullstrip. You can add an '-orig_vol' option in the command to get a good CSF mask.

Best,
Xi-Nian