search

jianpengz / DoDNet

[CVPR2021] DoDNet: Learning to segment multi-organ and tumors from multiple partially labeled datasets
GNU General Public License v3.0
177 stars 30 forks source link

Data download #1

Closed wanglixilinx closed 3 years ago

wanglixilinx commented 3 years ago

Hi. @jianpengz ,， Thx very much for your work. Currently, I just use this repro, and occur some problems when downloading the dataset as your instruction. 1.) I do not find any data related with Kidney dataset; 2) I just download Lits and found that are different with your data folder format in the lits/MOTS/MOTS_train/test.txt

jianpengz commented 3 years ago

Thanks for your attention.

  1. The link to the kidney dataset has been updated. Please try again.
  2. After downloading these datasets, put them in 'dataset/0123456/' and change the folder name, like '0Liver', '1Kidney'. Your data path should be consistent with the list file.
wanglixilinx commented 3 years ago

Hi， thanks for you reply. for the Liver dataset, my data folder is listed as below: train./Train-part1.zip train./Train-part2.zip train./Train-part3.zip train./Train-part4.zip train./Train-part5.zip train./Train-part6.zip

Is it ok ?

jianpengz commented 3 years ago

I think it should be like

WeChatd011ae95976ce1e5ccb33989ade8371d
wanglixilinx commented 3 years ago

Following your link, I download the data just as I listed above. Can you share your Liver dataset with me?

jianpengz commented 3 years ago

Please try to download via this Google drive link: [https://drive.google.com/drive/folders/0B0vscETPGI1-Q1h1WFdEM2FHSUE]. The items in this link should be like

image

arinaruck commented 3 years ago

Hello, @jianpengz! I'm dealing with the same issue right now: image

The image data is in the way you are listing it, but it is not following the LabelsTr, ImagesTr structure: image

How should the code be fixed for that case? And how exactly should the LiTS data be stored then? Like: 0Liver/Training Batch 1, 0Liver/Training Batch 2? Thanks

arinaruck commented 3 years ago

@jianpengz, would you also mind clarifying about the 1Kidney? Based on the GitHub, providing the data it is stored as follows: image Nevertheless, there is an if condition for the 1Kidney case, which is never satisfied, since there is no 'origin' in the data structure Maybe there is some additional step before re_spacing, which I am missing? I would also suggest rewriting the re_spacing code to multy processing pipeline, and I would be happy to do it as soon as the original preprocessing version works Thanks for your work

jianpengz commented 3 years ago

Hi @arinaruck, LiTS dataset has a different storage format from the MSD style. A simple solution is using the MSD style LiTS dataset from MSD challenge (both of them are the same). As for kidney dataset, I renamed the data/ folder as origin/ due to some reasons a long time ago. You can rename the download data/ as origin/ to work well.

arinaruck commented 3 years ago

@jianpengz Thank you!

arinaruck commented 3 years ago

@jianpengz, Sorry to bother you again, but shouldn't there be an else: or continue for the kidney case: https://github.com/jianpengz/DoDNet/blob/956d2558fc6e9e786260590d0e5c7843edd95f32/dataset/re_spacing.py#L28 Seems like the next for loop is for MSD only? I've just added a pull request fixing it if it indeed is the case. Please correct me, if I misinterpreted the code