Skip to content

Latest commit

 

History

History

augmentation

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
aug_phase.sh
aug_phase.sh
 
 
augment_phase.py
augment_phase.py
 
 
check_missing.sh
check_missing.sh
 
 
preprocess.py
preprocess.py
 
 
readme.md
readme.md
 
 
runner
runner
 
 

readme.md

Data preparation

Please go to the webpage for Autism Brain Imaging Data Exchange (ABIDE) and download the dataset that contains nii files. Before using program preprocess.py to preprocess them, make sure you have installed library spreco.

python preprocess.py --folder={the folder contains the nii files} --savepath={where you want to save preprocessed images}

Phase augmentation

Before performing phase augmentation, make sure you have downloaded models from this zenodo page

wget https://zenodo.org/record/6521188/files/models.tar

and make sure you create a list of files like below after the data preparation step

./dataset/abide_2/train/abide_1000000.npz
./dataset/abide_2/train/abide_1000001.npz
./dataset/abide_2/train/abide_1000002.npz
./dataset/abide_2/train/abide_1000003.npz

Then use program aug_phase.sh to start augmentation tasks. Before that, please check if the following things in phase_abide.yaml are correct

  1. specify where the downloaded model is with model_folder
  2. specify which model to use with model_name
  3. specify where to save the phase-augmented images with save_path

and the following things in aug_phase.sh

  1. specify the location of config file with CONFIG
  2. specify the location of the list of files with FILELIST
  3. the runner can parallelize your jobs, specify the number of GPU available with NUM_GPUS and GPU_IDS
  4. if your GPU has a not sufficient memory, reduce FILE_BATCHES
  5. specify your tmp_list with TMP