[BUG] If I have 2 runs of a task each in a different session and I select only one session (eg second one), preproc fails at coregistration

[marcobarilari]

Is there an existing issue for this?

• I have searched the existing issues

Operating system

• Linux
• Mac
• Windows

Operating system version

No response

SPM 12 version

• 7771
• 7487
• 7219

Platform

• MATLAB
• Octave

Platform version

No response

bidspm version

v3.1.0

bidspm branch / commit number

No response

Expected Behavior

To work with only 1 run even though 2 are present in the bids dataset

BIDS stats model

No response

Current Behavior & Error message

The batch is built in a way that expects 2 sessions at the coregistration level

Error:

21-Feb-2024 21:00:10 - Done    'Realign & Unwarp'
Dependency source available, but is missing required output.
Index exceeds matrix dimensions.
In item Other Images:
Dependency 1: Realign: Estimate & Reslice/Unwarp: Realigned Images (Sess 1) (out.sess(1).uwrfiles)
    '/Volumes/PAUL/datalad/analysis_high-res_BLAM_biMotion_localizers/outputs/derivatives/bidspm-preproc/sub-SC08/ses-02/func/usub-SC08_ses-02_task-auditoryLocalizer_run-02_part-mag_space-individual_desc-stc_bold.nii'


Dependency 2: Realign: Estimate & Reslice/Unwarp: Realigned Images (Sess 2) (out.sess(2).uwrfiles)
       sess: [1×1 struct]
    meanuwr: {'/Volumes/PAUL/datalad/analysis_high-res_BLAM_biMotion_localizers/outputs/derivatives/bidspm-preproc/sub-SC08/ses-02/func/meanusub-SC08_ses-02_task-auditoryLocalizer_run-02_part-mag_space-individual_desc-stc_bold.nii'}

Batch

here it is expecting 2 other images (2 runs but I selected only one through the session filter)

matlabbatch{3}.spm.spatial.coreg.estimate.other(1) = cfg_dep('Realign: Estimate & Reslice/Unwarp: Realigned Images (Sess 1)', substruct('.','val', '{}',{2}, '.','val', '{}',{1}, '.','val', '{}',{1}, '.','val', '{}',{1}), substruct('.','sess', '()',{1}, '.','uwrfiles'));
matlabbatch{3}.spm.spatial.coreg.estimate.other(2) = cfg_dep('Realign: Estimate & Reslice/Unwarp: Realigned Images (Sess 2)', substruct('.','val', '{}',{2}, '.','val', '{}',{1}, '.','val', '{}',{1}, '.','val', '{}',{1}), substruct('.','sess', '()',{2}, '.','uwrfiles'));

Anything else?

No rush, my workaround was to delete in raw the run I don't need (which is to discard anyway).

More pointers for solutions:

• Have a filter for run
• Be smarter that if a user select a session, metadata from other sessions are not considered in the batch

[github-actions]

We found the following entry in the FAQ which you may find helpful:

• General: Can I use parallelization with bidspm?

Feel free to close this issue if you found an answer in the FAQ. Otherwise, please give us a little time to review.

This is an automated reply, generated by FAQtory

[Remi-Gau]

does it also fail if you have 2 sessions but only select the first one?

[marcobarilari]

I don't know, I will give it a trial and let you know.

[marcobarilari]

Getting back to this issue, did not try yet but it is on my to do

[Remi-Gau]

I can reproduce the bug on one of the openneuro datasets

clear;
clc;

addpath(fullfile(pwd, '..', '..'));
bidspm();

participant_label = {'01'};
TASK = 'linebisection'; 

root_dir = fileparts(mfilename('fullpath'));
bids_dir = fullfile(root_dir, 'inputs', 'ds000114');
output_dir = fullfile(root_dir, 'outputs', 'ds000114', 'derivatives');
preproc = fullfile(output_dir, 'bidspm-preproc');

bidspm(bids_dir, output_dir, 'subject', ...
       'participant_label', participant_label, ...
       'action', 'copy', ...
       'task', TASK, ...
       'skip_validation', true, ...
       'verbosity', 3);

bids_filter_file = struct(...
      'bold', struct('modality', 'func', 'suffix', 'bold', 'ses', 'retest'), ...
      't1w',  struct('modality', 'anat', 'suffix', 'T1w'));
bidspm(bids_dir, output_dir, 'subject', ...
       'participant_label', participant_label, ...
       'action', 'preprocess', ...
       'task', TASK, ...
       'space', {'IXI549Space'}, ...
       'skip_validation', true, ...
       'bids_filter_file', bids_filter_file, ...
       'fwhm', 6, ...
       'verbosity', 3);

[Remi-Gau]

It seems that if you copy only a single session in from the raw dataset things work fine.