Processing issues in r20200830: GM and MTS segmentation

[valosekj]

Here, I list data quality issues related to r20200830 release.

I focused on outliers in generated figures by sq_generate_figures.py.

So far, I found problems with MT SC segmentations:

Bad SC segmentation sub-fslAchieva03_acq-T1w_MTS_seg.nii.gz

Bad SC segmentation sub-fslAchieva04_acq-T1w_MTS_seg.nii.gz

Bad SC segmentation sub-fslAchieva05_acq-T1w_MTS_seg.nii.gz

Bad SC segmentation sub-fslAchieva06_acq-T1w_MTS_seg.nii.gz

And GM segmentation issues:

Bad GM segmentation sub-fslAchieva04_T2star_rms_gmseg.nii.gz

Missing GM segmentation in a few slices sub-tokyoIngenia04_T2star_rms_gmseg.nii.gz

[evaalonsoortiz]

From the figures I generated, I identified the following outliers:

MTR: geneva02

QC images do not show any obvious segmentation or labeling errors

CSA-T2w: sub-oxfordFmrib04

QC images do not show any obvious segmentation or labeling errors

[jcohenadad]

CSA-T2w: sub-oxfordFmrib04: the subject was supposed to be excluded from the stats: https://github.com/spine-generic/data-multi-subject/blob/master/exclude.yml#L5

@evaalonsoortiz was it included in the figure generation (in that case it's a bug)?

same thing for geneva02/mtr: https://github.com/spine-generic/data-multi-subject/blob/master/exclude.yml#L14

[jcohenadad]

Bad GM segmentation sub-fslAchieva04_T2star_rms_gmseg.nii.gz

--> need to exclude

[jcohenadad]

@valosekj @evaalonsoortiz can you please move the list of data quality issue to the specific issue with respective contrast (create one if does not exist). The purpose of these "data-quality" issue are to gather all problems with data quality. It is not specific to a release (ie no relation with the processing). However, if there are new subjects to exclude for the analysis/figure, please start a branch to modify the exclude.yml file. Thanks!

[evaalonsoortiz]

CSA-T2w: sub-oxfordFmrib04: the subject was supposed to be excluded from the stats: https://github.com/spine-generic/data-multi-subject/blob/master/exclude.yml#L5

@evaalonsoortiz was it included in the figure generation (in that case it's a bug)?

same thing for geneva02/mtr: https://github.com/spine-generic/data-multi-subject/blob/master/exclude.yml#L14

Ah, no my bad. I did not exclude any subjects, I had assumed those were excluded when the processing step is taken.

[jcohenadad]

GM seg to fix:

[jcohenadad]

GM segmentation fixed in https://github.com/spine-generic/data-multi-subject/pull/40

[jcohenadad]

MTS segmentations fixed in #43