Poor data quality: PSIR

[valosekj]

This issue summarizes PSIR images with poor data quality.

• sub-mon118_ses-M0_PSIR.nii.gz - severe motion artifact (original issue here) - added to exclude.yml

[plbenveniste]

Similar to sub-mon118 Similarly to the above subject (sub_mon118), the following files were found to have similar issue of data quality:

• sub_mon009_ses-M0_PSIR.nii.gz
• sub_mon032_ses-M0_PSIR.nii.gz
• sub_mon097_ses-M0_PSIR.nii.gz
• sub-mon148_ses-M0_PSIR.nii.gz
• sub-mon168_ses-M0_PSIR.nii.gz
• sub-mon191_ses-M0_PSIR.nii.gz
• sub-van176_ses-M0_PSIR.nii.gz

In the above images, we can see the shift from black to white, from one slide to another

Blurry image The following subject has a very blurry image:

• sub-van206_ses-M0_PSIR.nii.gz Should it be disccarded and included in the exlude.yml file ? GIF of sub-van206

Motion The following subject has very strong motion on alternating slices :

• sub-tor133_ses-M0_PSIR.nii.gz Should it be discarded ? GIF of sub-tor133

[jcohenadad]

sub-tor133_ses-M0_PSIR.nii.gz: Should it be discarded ?

Yes I would say so

p.s. I suggest to create one post per subject (instead of showing multiple subjects per post), so that we can more easily cross-ref one subject with a unique URL (eg: in the exclude.yml file, when communicating with Shannon, etc.)

[valosekj]

sub-van206_ses-M0_PSIR.nii.gz Should it be disccarded and included in the exlude.yml file ?

Yes, please, include the image to the exlude.yml file

sub-tor133_ses-M0_PSIR.nii.gz Should it be discarded ?

Yes, please, include the image to the exlude.yml file

[plbenveniste]

Referencing a data quality issue for sub-mon006_ses-M0 which was previously described in this issue

• sub-mon006 - has FOV covering the whole brain. The SC has only C1-C6

[plbenveniste]

Problem with sub_mon009_ses-M0_PSIR.nii.gz. Linked to a strong motion artifact.

Similar image quality problem as in issue 43 for image sub-mon118_ses-M0

Image added to exclude.yml to exclude it in future works.

[plbenveniste]

Problem with sub_mon032_ses-M0_PSIR.nii.gz. Linked to a strong motion artifact.

Similar image quality problem as in issue 43 for image sub-mon118_ses-M0

Image added to exclude.yml to exclude it in future works.

[plbenveniste]

Problem with sub_mon097_ses-M0_PSIR.nii.gz. Linked to a strong motion artifact.

Similar image quality problem as in issue 43 for image sub-mon118_ses-M0

Image added to exclude.yml to exclude it in future works.

[plbenveniste]

Problem with sub_mon148_ses-M0_PSIR.nii.gz. Linked to a strong motion artifact.

Similar image quality problem as in issue 43 for image sub-mon118_ses-M0

Image added to exclude.yml to exclude it in future works.

[plbenveniste]

Problem with sub_mon168_ses-M0_PSIR.nii.gz. Linked to a strong motion artifact.

Similar image quality problem as in issue 43 for image sub-mon118_ses-M0

Image added to exclude.yml to exclude it in future works.

[plbenveniste]

Problem with sub_mon191_ses-M0_PSIR.nii.gz. Linked to a strong motion artifact.

Similar image quality problem as in issue 43 for image sub-mon118_ses-M0

Image added to exclude.yml to exclude it in future works.

[plbenveniste]

Problem with sub_van176_ses-M0_PSIR.nii.gz. Linked to a strong motion artifact.

Similar image quality problem as in issue 43 for image sub-mon118_ses-M0

Image added to exclude.yml to exclude it in future works.

[plbenveniste]

Problem with sub_van206_ses-M0_PSIR.nii.gz. The image is very blury as it can be seen in the below GIF:

Image added to exclude.yml to exclude it in future works.

[plbenveniste]

Problem with sub_tor133_ses-M0_PSIR.nii.gz. There is a very big motion which can be seen on alternating slices. It can be seen in the below GIF:

Image added to exclude.yml to exclude it in future works.

[valosekj]

• sub-tor090_ses-M0_PSIR.nii.gz - strong motion artifact. SC seg apparently works, but disc labeling is tricky - keeping for now:

[valosekj]

• sub-tor092_ses-M0_PSIR.nii.gz - motion between slices - SC still visible - keeping for now:

[plbenveniste]

Poor image quality reported while doing QC in this post (link):

• sub-mon113_ses-M0

• [x] Need to exclude it : and add to exclude.yml

[plbenveniste]

Poor image quality reported while doing QC in this post (link):

• sub-mon138_ses-M0

The quality is poor, but the spinal cord is still visible. Keeping it for now

[plbenveniste]

Poor image quality reported while doing QC in this post (link):

• sub-mon152_ses-M0

• [x] Need to exclude it : and add to exclude.yml

[plbenveniste]

Poor image quality reported while doing QC in this post (link): motion on alternating slices

• sub-van156_ses-M0

SC segmentation didn't work well, but SC is pretty visible: keeping it for now

[jcohenadad]

p.s. for all the subjects showing motion across odd/even slices, there is a "simple" pre-processing method we could implement, which consists in co-registering the odd and even slices.

[valosekj]

p.s. for all the subjects showing motion across odd/even slices, there is a "simple" pre-processing method we could implement, which consists in co-registering the odd and even slices.

Good point! Cross-referring relevant issue: https://github.com/ivadomed/canproco/issues/19

[plbenveniste]

sub-mon175_ses-M0 has poor image quality

Keeping it for now as SC is still visible

[plbenveniste]

Subject sub-tor014 should be removed : too blury

• [x] add to exclude.yml file

[valosekj]

• sub-van134_ses-M0_PSIR.nii.gz imposible to perform disc labeling:

Keeping it for now as SC is still visible

[valosekj]

• sub-van135_ses-M0_PSIR.nii.gz imposible to perform disc labeling:

Keeping it for now as SC is still visible

[valosekj]

• sub-van154_ses-M0_PSIR.nii.gz - motion between slices - SC still visible - keeping for now:

[valosekj]

• sub-van171_ses-M0_PSIR.nii.gz imposible to perform disc labeling:

Keeping it for now as SC is still visible

[valosekj]

• sub-van207_ses-M0_PSIR.nii.gz imposible to perform disc labeling:

Keeping it for now as SC is still visible

[plbenveniste]

Impossible to perform disc_labeling on subject sub-van181_ses-M0_PSIR.nii.gz

Keeping it as SC is still visible.

[plbenveniste]

Impossible to perform disc_labeling on subject sub-van201_ses-M0_PSIR.nii.gz

[plbenveniste]

Subject sub-mon050_ses-M0_PSIR has strong motion : added to exclude.yml file

[plbenveniste]

Small motion artifact on sub-mon118_ses-M12 making lesion detection/segmentation too complicated

[simonqueric]

Motion artefact on sub-mon077_ses-M0.

Should we add it to the exclude.yml ?

[plbenveniste]

Motion artefact on sub-mon077_ses-M0. Should we add it to the exclude.yml ?

Yes, I think so!

[simonqueric]

Small motion artefact on sub-mon029_ses-M0 :

[plbenveniste]

Small motion artefact on sub-mon029_ses-M0 :

I added this one and the previous one as well.

[simonqueric]

• Motion artefact on sub-mon104_ses-M0

[plbenveniste]

Good catch! Added as well!

[simonqueric]

• Small motion artefact on sub-mon001_ses-M0

As we can see the spinal cord, I'm not sure if we should put it on exclude.yml.

[plbenveniste]

Hum, that's a good question! I would suggest not adding it to the exclude.yml since the image is good enough to obtain both a SC seg and a lesion seg (and the subject has lesions). What do you think @valosekj @jcohenadad ?

[jcohenadad]

I would suggest not adding it to the exclude.yml since the image is good enough to obtain both a SC seg and a lesion seg (and the subject has lesions).

agreed

[simonqueric]

• Motion artefact on sub-mon036_ses-M0 :

The segmentation of the spinal cord seems alright.

[simonqueric]

• Motion artefact on sub-mon055_ses-M12 :

The segmentation is bad. Added to exclude.yml

[simonqueric]

• Strong motion artefact on sub-mon150_ses-M12 :

Added to exclude.yml

[simonqueric]

• Motion artefact on sub-mon185_ses-M12

The segmentation fails but the lesions are visible. Keeping it for now.

[simonqueric]

• Motion artefact on sub-mon034_ses-M0

The segmentation of the spinal cord seems alright.

[simonqueric]

• Stron motion artefact on sub-mon096_ses-M12

Added to exclude.yml

[plbenveniste]

Referencing a previous file sub-mon052_ses-M0 which has already been added : link