search

Mathilde-16 / STOP_MS_data

0 stars 0 forks source link

Flag a QC with "Fail" or "Pass" in order to set up the manual segmentation #2

Closed Mathilde-16 closed 3 months ago

Mathilde-16 commented 5 months ago

Regarding the manual segmentation, I haven't done it yet because I would like to know, please, if you can help me find out how to determine whether segmentation worked or not when we look at the QC (in order to flag it with “Fail” or “Pass”) ?

I guess isn't always easy to spot the issues and that it's very important to get this right and to make no mistakes.

Thank you for your help !

Mathilde-16 commented 5 months ago

Hi again, just a few additional details !

The contrast agnostic segmentation qc (index.html file) shows better results than the segmentation qc, as we could have expected. Indeed, output for the sct_deepseg_sc command looks like this for several people : Screenshot.pdf

I tried to flag and complete both QC corresponding to the links below. I've probably made several mistakes but it was just a training exercise before your recommendations.

jcohenadad commented 5 months ago

Regarding the manual segmentation, I haven't done it yet because I would like to know, please, if you can help me find out how to determine whether segmentation worked or not when we look at the QC (in order to flag it with “Fail” or “Pass”) ?

i suggest you go through a QC, flag pass/fail, and i can have a quick look to let you know if i agree with it

Mathilde-16 commented 5 months ago

I tried to flag and complete both QC corresponding to the links below. I've probably made several mistakes but it was just a training exercise before your recommendations.

segmentation with sct_deepseg_sc, and labeling : file:///Users/mathilde/Desktop/BIDS_DATASET_seg_copy_3_seg+label/sct_v6.3/sct_v6.3_segmentation/OUTPUT/qc/index.html

contrast agnostic segmentation, and labeling : file:///Users/mathilde/Desktop/BIDS_DATASET_contrast-agnost-seg+label_copy_4/derivaties/sct_v6.3_contrast-agnostic/qc/index.html

I think what I did above is what you mean, no ? If you copy and paste both links into a search engine, you should have access to the QCs I tried to evaluate ?

jcohenadad commented 5 months ago

no-- you pointed to a LOCAL URL https://askubuntu.com/questions/932215/make-url-point-to-local-folder

check your URL: it starts with "local"-- you need to downlaod, zip, upload here (or somewhere else on the cloud if too big)

Mathilde-16 commented 5 months ago

Please find below both QC

jcohenadad commented 5 months ago

in https://github.com/Mathilde-16/STOP_MS_data/issues/2#issuecomment-2138237283 you only zipped the HTML file but not the images that the HTML code depends on-- you need to zip the entire QC folder--

Mathilde-16 commented 5 months ago

My apologies.

Please find below both compressed QC folders,

jcohenadad commented 5 months ago

The contrast agnostic segmentation qc (index.html file) shows better results than the segmentation qc, as we could have expected

Agreed!

jcohenadad commented 3 months ago

we can consider this issue fixed-- closing