LiTs Challenege Semantic Segmentation of Liver from CT Scans
-
Clone The Repository https://github.com/imatge-upc/liverseg-2017-nipsws
Put in inside the current directory as shown here. Replace the 'seg_liver.py', 'seg_liver_train.py', seg_liver.py' files inside the cloned repository by the files provided here. -
Create a Folder
- LiTS_database inside liverseg folder.
- create a results folder insider liverseg folder
- Download the weights from here and add them to train_files folder.
-
Process NIFTI Files
python process_test_database.py dataset_folder_name
Input - ‘volume_i_.nii’ files from the dataset_folder
Output - ‘/LiTS_database/test_image_volumes/’ -
Create a File Containing Path to Test Images
python Create_test.py test_image_volumes
Input - "liverseg-2017-nipsws/LiTS_database/folder_name" Output - liverseg-2017-nipsws/seg_DatasetList/test.txt -
Test the Trained Model Download the weights from here
python liverseg-2017-nipsws/seg_liver_test.py
Input - ‘seg_DatasetList/test.txt'
Output - ‘liverseg-2017-nipsws/results’
Convert Multile Images Slices into a 3D Volume
python npy_2_volume.py
Converts multiple .npy output slices into a single .npy volume for a particular case.
Input - "liverseg-2017-nipsws/results/seg_liver"
Output - “liverseg-2017-nipsws/results/output_volumes/”