Lesion-Aware Dynamic Kernel for Polyp Segmentation

Introduction

This repository contains the PyTorch implementation of:

Lesion-Aware Dynamic Kernel for Polyp Segmentation, MICCAI 2022.

Requirements

• torch
• torchvision
• tqdm
• opencv
• scipy
• skimage
• PIL
• numpy

Usage

1. Training

python train.py  --root /path-to-project  --mode train
--train_data_dir /path-to-train_data   --valid_data_dir  /path-to-valid_data

2. Inference

python test.py  --root /path-to-project  --mode test  --load_ckpt checkpoint  
--test_data_dir  /path-to-test_data

Citation

If you feel this work is helpful, please cite our paper

@inproceedings{zhang2022lesion,
  title={Lesion-Aware Dynamic Kernel for Polyp Segmentation},
  author={Zhang, Ruifei and Lai, Peiwen and Wan, Xiang and Fan, De-Jun and Gao, Feng and Wu, Xiao-Jian and Li, Guanbin},
  booktitle={International Conference on Medical Image Computing and Computer-Assisted Intervention},
  pages={99--109},
  year={2022},
  organization={Springer}
}