aces / cbrain-plugins-fsl Goto Github PK

Step 1: QC based on registration only.
Step 2: QC based on whether the components are noise.

Increase flexibility in naming FSL outputs, similarly to CIVET.

• show the registration options only when registration checkbox is ticked.
• control nonlinear warp field resolution should appear only if nonlinear registration is selected
• if the user chooses the automatic detection, then dim the choice box for entering values.
• re-evaluate all the warnings when value changes.
• if we select analysis=1 (type of analysis = 1), then hide the stats section.
• show standard image selector only if "use alternate reference image?" is ticked.
• show "output components" only if "automatic dimensionality estimation" is not selected.

Need to start from previously pre-processed data.

For instance, "command not found" error is not handled because:

• It doesn't write ERROR in stdout/stderr.
• The output directory exists (it's empty) and can be transferred.

Look at how it's done in CIVET.

in TICA & GICA, make sure all input files have the same dimension. If not, stop the task, create a report of filenames and their dimensions and the user will be instructed to go to the parameter file and de-select the violating files.

MELODIC input files should include at least one set of anatomical and one set of 4D fMRI data. The 4D fMRI data should have a dimension of (fslnvols - ndelete)> 20. Otherwise, the CBRAIN task should not proceed, and the user should get a red-flag warning.

Create a test base as follows.

Data:

• two subjects in nifti and in minc, with identical dimensions and TRs (DONE)
• the corresponding 2 "correct" CSV files (nifti and minc) (DONE)
• two subjects in nifti, with different dimensions (DONE
• two subjects in nifti, with different TRs (DONE)
• two subjects in nifti, with less than 22 volumes (DONE)
• the corresponding 3 "incorrect" CSV files (different dimensions and different TRs)
• a custom regstandard file in nifti and in minc. (DONE)
• a design file, with ndelete=5
  This data should be put in a separate git repo "cbrain-test-melodic". Ultimately, a script could be built that registers this data in CBRAIN and launches the tests through the API.

Tests with the 2 "correct" CSV files:

1. individual analyses in minc, no auto-correction, no custom registration file
2. individual analyses in nifti, no auto-correction, no custom registration file
3. individual analyses in nifti, no auto-correction, with custom registration file in nifti
4. individual analyses in nifti, no auto-correction, with custom registration file in minc
5. individual analyses in nifti, with auto-correction, no custom registration file
6. group analyses in nifti
7. group analyses in minc

Test passes iif:

• task(s) are completed
• result files have correct parents
• viewer works on result files
• QC framework works

Tests with the 3 "incorrect" CSV files:
When subjects have more than 22 volumes:

1. individual analyses: tasks must complete .
2. group analyses: tasks must fail and report the proper error message.
   When subjects have less than 22 volumes:

• All tasks must fail with proper error message.

BET can run on 4D fMRI data by selecting option -F. If this option is selected, other ones, such as -B, -S, -R must become unwritable. Otherwise, the output will be seriously erroneous, and only one volume.

Imported from Redmine ticket #7771.

• tr can be extracted with fslhd
• npts can be extracted with fslnvols

It is not a good idea to do it on the portal side, as (1) it will make the portal installation dependent on FSL tools, (2) running command-line tools from the portal code can slow down the portal interface. So these adjustments have to be made on the clusters, which means that users will not be able to check/modify these values.

Najma said: we should just provide it as an option for the user to set up on the interface, with a warning that the values provided through the interface will override the values that may have been provided by the user in the fsf file.

individual files are generated outside the GICA directory. Gather them all under the higher_level analysis. modify the viewer for GICA accordingly.

The QC on these files should be done the same as in individual ICA (already implemented).

Add a selection list containing user files of type "template". See how it is done in randomize.

• Select "Registration to main structural" if it was not enabled in the design file.
• Select "Use alternate reference image?" when alternate file is provided and this option is not enabled.
• Select "Registration to standard image?" when group analysis is performed?

Print a warning each time a parameter is adjusted.