Skip to content

Latest commit

 

History

History
36 lines (28 loc) · 814 Bytes

README.md

File metadata and controls

36 lines (28 loc) · 814 Bytes

validating-cnn-mgmt

Requirements

pyyaml
tqdm
numpy
pandas
dicom2nifti
antspyx
sklearn
torch
torchvision
torchio
monai

How to reproduce

  1. Download the BraTS dataset. You must be registered and logged in to Kaggle.

  2. Extract the zip file to the desired path. The extracted path should be passed to the --source_dir option in the preprocessing step.

  3. Run the preprocessing step for both train and test datasets:

python preprocess.py --source_dir path_to_brats_data --split train
python preprocess.py --source_dir path_to_brats_data --split test
  1. Run the validation experiments:
python validate_sweep.py 
  • To customize the validation experiment sweep, modify config.yaml.