Repository files navigation

I used these scripts to convert thePROSTATEx-Seg-HiResDICOM dataset to a more friendly nii format. This dataset is not straightforward to convert for various reasons,for example, you can usedcm2niix to convert the mri images but not thesegmentations, there isn't a one-to-one relation between segmentation slices and mri slices and so on.Remember: this code is far from perfect, it's not well engineered, and it was written for personal necessity.For a much cleaner code with automated downloads of data and tools seethis gist.

1. Download the dataset with classic directory names option (both mri and segmentations).

2. Create a new folder (for example named ProstateX) with two sub-folders in it, one for the mri images and one for the segmentations.

3. Place all the folders (ProstateX-0004, ecc...) in the respective sub-folder.Do this for both the mri images and the segmentations. Your folder structure should look like this (truncated for clarity):

   ProstateX├───mri│   ├───ProstateX-0004│   ├───ProstateX-0007│   ├───ProstateX-0009│   └───ProstateX-...└───seg    ├───ProstateX-0004    ├───ProstateX-0007    ├───ProstateX-0009    └───ProstateX-...

4. Run rename_all_folders.py two times: the first time thedcm_path variableshould point to the mri sub-folder, the second time it should point to the segmentations sub-folder.

5. Download thedcm2niix executable (dcm2niix.exe) and store itwhere you can use it.

6. Run mri2nii.py adjusting the various paths in it.dcm_dataset_path shouldpoint to the mri sub-folder,nii_dataset_path to your preferred output folderandconverter to dcm2niix.exe.

7. Run seg2nii.py adjusting the various paths in it.dcm_dataset_path shouldpoint to the segmentations sub-folder, andnii_dataset_path should pointto the same location as before.

8. Enjoy the converted dataset!