ivadomed / canproco

Code for preprocessing the CanProCo brain and spinal cord dataset
MIT License
4 stars 1 forks source link

Manual segmentation of 3 missing M0 missing lesion mask #99

Open plbenveniste opened 1 month ago

plbenveniste commented 1 month ago

Opening this issue thanks to a comment from @leelisae regarding some missing lesion mask at M0.

Related issue #96

On it now with the following config file:

FILES_LESION:
- sub-edm161_ses-M0_PSIR.nii.gz
- sub-mon015_ses-M0_PSIR.nii.gz
- sub-mon052_ses-M0_PSIR.nii.gz
- sub-mon164_ses-M0_PSIR.nii.gz

Running the following code to perform manual segmentation of MS lesions:

python manual_correction.py -path-img ~/tmp_romane/ms_lesion_agnostic/data/canproco  -config ~/Desktop/manual_seg_lesion_missing_canproco.yaml  -path-label ~/tmp_romane/ms_lesion_agnostic/data/canproco/derivatives/labels  -suffix-files-lesion _lesion-manual -fsleyes-dr="-40,70"
plbenveniste commented 1 month ago

I think that the mask for sub-mon052_ses-M0_PSIR.nii.gz should be empty, and the image added to the exclude file. I think that the spots that look like lesions are actually some artifacts.

GIF related: ezgif com-animated-gif-maker (14)

plbenveniste commented 1 month ago

The subject sub-mon052_ses-M0 had already been added to the exclude.yml file: here. Therefore, the mask is empty.

All segmentations were pushed to branch plb/segment_missing_4_lesion_seg

@jcohenadad Could you review the 3 segmentations before I send them to @leelisae ?

Also, @leelisae do you take into account the exclude.yml file when doing your computations ? We reference all the files which are problematic and which we don't use in our work. Usually the segmentations files are empty for them.

jcohenadad commented 1 month ago

@jcohenadad Could you review the 3 segmentations before I send them to @leelisae ?

I cannot because of https://github.com/neuropoly/data-management/issues/321

If you'd like me to review ASAP, zip the files and upload them here (these are binary files-- it will be <1MB once zipped)

plbenveniste commented 1 month ago

Here are the 3 manually segmented segmentation file: manual_segs.zip

jcohenadad commented 1 month ago

hum... i don't seem to have M0:

image

and git pull does not work https://github.com/neuropoly/data-management/issues/321 (@namgo )

can you email me the files?

jcohenadad commented 1 month ago

Great job @plbenveniste ! Here are my modifications (I only edited two subjects, but I modified the three JSON files to record the fact that I have reviewed the three segmentations): manual_segs.zip

plbenveniste commented 1 month ago

Thanks a lot @jcohenadad !

Segmentations done and added to branch plb/segment_missing_4_lesion_seg. PR was created and waiting for review.

plbenveniste commented 1 month ago

Here it is: exclude.yml (it's stored in this repo).