/Automated-and-Editable-Prompt-Learning-for-Brain-Tumor-Segmentation

Code for the proposed Automated and Editable Prompt Learning for Brain Tumor Segmentation

Primary LanguagePythonApache License 2.0Apache-2.0

AEPL: Automated and Editable Prompt Learning for Brain Tumor Segmentation

This repository provides the code for "AEPL: Automated and Editable Prompt Learning for Brain Tumor Segmentation".

AEPL Fig. 1. Structure of AEPL.

Requirementss

Some important required packages include:

  • Pytorch version >=0.4.1.
  • Python == 3.7
  • Some basic python packages such as Numpy.
  • nnUNetv1
  • Follow official guidance to install Pytorch.
  • Follow official guidance to install nnUNetv1.

Dataset

You experiment on BraTS2018.

Usages

Preprocess

python ./preprocess/BraTS.py
python ./preprocess/generate_json.py
nnUNet_plan_and_preprocess -t 501 --verify_dataset_integrity

Train

CUDA_VISIBLE_DEVICES=0 nnUNet_train 3d_fullres nnUNetTrainerV2_AEPL 501 0

Test

CUDA_VISIBLE_DEVICES=0 nnUNet_predict -i /nnUNet/nnUNet_raw/nnUNet_raw_data/Task501_BraTS/imagesTs/ -o /nnUNet/nnUNet_output/Task501_BraTS/nnUNetTrainerV2_AEPL/fold_0 -t 501 -m 3d_fullres -f 0 -tr nnUNetTrainerV2_AEPL
  1. Our experimental results are shown in the table: refinement