README.md

[CBM2024]DAC-Net

🔥🔥This repo is the official implementation of 'DAC-Net : An Light-weight U-shaped Network Based Efficient Convolution And Dual-Attention for Thyroid Nodule Segmentation', accepted at Computers in Biology and Medicine2024🔥🔥

Main Environments

• python 3.9
• pytorch 2.1.0
• torchvision 0.16.0

Requirements

Install from the requirements.txt using:

pip install -r requirements.txt

Prepare the dataset.

• The DDTI datasets can be found here (Link) and the TN3K datasets can be found here (Link), divided into a 7:1:2 ratio.

• Then prepare the datasets in the following format for easy use of the code:

├── datasets
    ├── DDTI
    │   ├── Test_Folder
    │   │   ├── img
    │   │   └── labelcol
    │   ├── Train_Folder
    │   │   ├── img
    │   │   └── labelcol
    │   └── Val_Folder
    │       ├── img
    │       └── labelcol
    └── TN3k
        ├── Test_Folder
        │   ├── img
        │   └── labelcol
        ├── Train_Folder
        │   ├── img
        │   └── labelcol
        └── Val_Folder
            ├── img
            └── labelcol

Train the Model

First, modify the model, dataset and training hyperparameters (including learning rate, batch size img size and optimizer etc) in Config.py

Then simply run the training code.

python train_model.py

Reference

• UNet
• UNet++
• UNet3+
• MultiResUNet
• TransUNet
• UCTransNet
• ACC_UNet

Citation

If you find this work useful in your research or use this dataset in your work, please consider citing the following papers:

@article{2024DAC,
  title={DAC-Net: A light-weight U-shaped network based efficient convolution and attention for thyroid nodule segmentation},
  author={ Yang, Yingwei  and  Huang, Haiguang  and  Shao, Yingsheng  and  Chen, Beilei },
  journal={Computers in Biology and Medicine},
  volume={180},
  year={2024},
}