NestedFormer

NestedFormer (MICCAI2022) is a multimodal segmentation model on 3D medical images. The features of different modalities are fused through tri-oriented self-attention and cross-attention. We also improve the Poolfromer structure (CVPR2022) as an efficient encoder.

Read our paper https://arxiv.org/abs/2208.14876 on ArXiv for a formal introduction.

Getting Started

Setup

pip install monai
pip install tqdm
pip install tensorboardX

Download data

Please download the brats2020 datasets. Of course, switching to other datasets is ok.

Run

python main.py --distributed  --logdir=log_train_nestedformer --fold=0 --json_list=./brats2020_datajson.json --max_epochs=1000 --lrschedule=warmup_cosine --val_every=10 --data_dir=/data/MICCAI_BraTS2020_TrainingData/  --out_channels=3 --batch_size=1 --infer_overlap=0.5

--data_dir is the location of the data.

Train your own dataset

The data processing code is in utils/data_utils.py. You can modify this code for your own dataset.

Acknowledgment

Our implementation is mainly based on the following codebases. We gratefully thank the authors for their wonderful works.

pytorch, monai, monai-research-contributions