[RESOLVED] Nonconformant arguments in extractNuisanceSignal resulting in error

Issues
1

Description:

Hi, I have [minimally] preprocessed my data hcp1-5, and have successfully performed mapHCPData, createBOLDBrainMasks, computeBOLDStats, createStatsReport. I am experiencing an error with extractNuisanceSignals. I suspect it’s due to the BOLDBrainMask and aparc+aseg having non-compatible voxel dimensions. BOLDBrainMask is 3mm isotropic (which is the correct sampling of this data set), while aparc+aseg_bold.nii.gz is 2mm isotropic (not sure where this parameter is specified).
Best,
Taku

Call:

qunexContainer extractNuisanceSignal \
    --sessionsfolder="$my_study_folder/sessions" \
    --sessions="$batch_file" \
    --overwrite='yes' \
    --container="$qunex_container"

Logs:

Running external command via Qu|Nex:

octave -q
–no-init-file
–eval “try g_ExtractNuisance(’/gpfs/loomis/pi/n3/Studies/MurrayLab/taku/multiTaskVAE/qunexMultiTaskVAE/sessions/03_b1/images/functional/bold8.nii.gz’, ‘/gpfs/loomis/pi/n3/Studies/MurrayLab/taku/multiTaskVAE/qunexMultiTaskVAE/sessions/03_b1/images/segmentation/freesurfer/mri/aparc+aseg_bold.nii.gz’, ‘/gpfs/loomis/pi/n3/Studies/MurrayLab/taku/multiTaskVAE/qunexMultiTaskVAE/sessions/03_b1/images/segmentation/boldmasks/bold8_frame1_brain_mask.nii.gz’, ‘/gpfs/loomis/pi/n3/Studies/MurrayLab/taku/multiTaskVAE/qunexMultiTaskVAE/sessions/03_b1/images/functional/movement’, ‘/gpfs/loomis/pi/n3/Studies/MurrayLab/taku/multiTaskVAE/qunexMultiTaskVAE/sessions/03_b1/images/ROI/nuisance’, ‘’, ‘’, ‘’, true, true); catch ME, g_ReportError(ME); exit(1), end; exit”

Test file: /gpfs/loomis/pi/n3/Studies/MurrayLab/taku/multiTaskVAE/qunexMultiTaskVAE/sessions/03_b1/images/functional/movement/bold8.nuisance

Running g_ExtractNuisance

Parameters:

        img: /gpfs/loomis/pi/n3/Studies/MurrayLab/taku/multiTaskVAE/qunexMultiTaskVAE/sessions/03_b1/images/functional/bold8.nii.gz
      fsimg: /gpfs/loomis/pi/n3/Studies/MurrayLab/taku/multiTaskVAE/qunexMultiTaskVAE/sessions/03_b1/images/segmentation/freesurfer/mri/aparc+aseg_bold.nii.gz
      bmimg: /gpfs/loomis/pi/n3/Studies/MurrayLab/taku/multiTaskVAE/qunexMultiTaskVAE/sessions/03_b1/images/segmentation/boldmasks/bold8_frame1_brain_mask.nii.gz
     target: /gpfs/loomis/pi/n3/Studies/MurrayLab/taku/multiTaskVAE/qunexMultiTaskVAE/sessions/03_b1/images/functional/movement
    ntarget: /gpfs/loomis/pi/n3/Studies/MurrayLab/taku/multiTaskVAE/qunexMultiTaskVAE/sessions/03_b1/images/ROI/nuisance
     wbmask: 
 sessionroi: 
       nroi: 
     shrink: 1

—> Reading bold [/gpfs/loomis/pi/n3/Studies/MurrayLab/taku/multiTaskVAE/qunexMultiTaskVAE/sessions/03_b1/images/functional/bold8.nii.gz]
—> Reading bold brain mask [/gpfs/loomis/pi/n3/Studies/MurrayLab/taku/multiTaskVAE/qunexMultiTaskVAE/sessions/03_b1/images/segmentation/boldmasks/bold8_frame1_brain_mask.nii.gz]
—> Reading segmentation mask [/gpfs/loomis/pi/n3/Studies/MurrayLab/taku/multiTaskVAE/qunexMultiTaskVAE/sessions/03_b1/images/segmentation/freesurfer/mri/aparc+aseg_bold.nii.gz]

=========================================
Execution error! Processing failed!
Please check arguments and/or try running the command in Matlab or Octave directly.

The exact error reported:

Error identifier: Octave:nonconformant-args
Error message: mx_el_and: nonconformant arguments (op1 is 259200x1, op2 is 902629x1)
Error stack: /opt/qunex/nitools/general/g_ExtractNuisance.m → g_ExtractNuisance [line: 117]

=========================================

2

I should mention that afterwards, a quick resampling of the aparc+aseg_bold.nii.gz file appeared to remedy the situation.

This resample was done using AFNI, though can also be done in FSL:

3dresample -input /gpfs/loomis/pi/n3/Studies/MurrayLab/taku/multiTaskVAE/qunexMultiTaskVAE/sessions/03_b1/images/segmentation/freesurfer/mri/aparc+aseg_bold.nii.gz -prefix /gpfs/loomis/pi/n3/Studies/MurrayLab/taku/multiTaskVAE/qunexMultiTaskVAE/sessions/03_b1/images/segmentation/freesurfer/mri/aparc+aseg_bold_2mm.nii.gz -master /gpfs/loomis/pi/n3/Studies/MurrayLab/taku/multiTaskVAE/qunexMultiTaskVAE/sessions/03_b1/images/segmentation/boldmasks/bold8_frame1_brain_mask.nii.gz

3

Hi Taku, we will assign a developer to this issue and will let you know once it has been resolved.

4

Taku, hi!

Indeed, irrespective of the BOLD voxel size in the dataset, the default and the expected voxels size for the processed volume BOLD images is 2 x 2 x 2 mm. This is the resolution used to prepare aseg+aparc.nii.gz files. I have opened a developer ticket so that aseg+aparc.nii.gz files will be resampled to the resolution specified by the hcp_bold_res and so match the preprocessed volume BOLD files. Before the issue is resolved in code, either manually resample aseg+aparc.nii.gz files in <session_id>/images/segmentation/freesurfer/mri or use the default 2mm setting for hcp_bold_res when processing BOLD images using HCP pipelines. To be able to use the different atlases and other support files for volume images, I would encourage you to use the default 2mm resolution.

With kind regards,

Grega