2D Deformable LKA

Instructions for the 2D version of the D-LKA net.

Model weights

You can download the learned weights of the D-LKA Net in the following table.

Task	Dataset	Learned weights
Multi organ segmentation	Synapse	D-LKA Net 2D
Skin 2017	Skin Dataset	D-LKA Net TODO
Skin 2018	Skin Dataset	D-LKA Net TODO
PH2	Skin Dataset	D-LKA Net TODO

Environment Setup

1. Create a new conda environment with python version 3.8.16:

   conda create -n "d_lka_net_2d" python=3.8.16
   conda activate d_lka_net_2d
   

2. Install PyTorch and torchvision

   pip install torch==1.11.0+cu113 torchvision==0.12.0+cu113 torchaudio==0.11.0+cu113 --extra-index-url https://download.pytorch.org/whl/cu113
   

3. Install the requirements with:

   pip install -r requirements.txt
   

Synapse Dataset

Training and Testing

1. Download the Synapse dataset from the link above.

2. Run the code below to train D-LKA Net on the Synapse dataset.

   python train_MaxViT_deform_LKA.py --root_path ./data/Synapse/train_npz --test_path ./data/Synapse/test_vol_h5 --batch_size 20 --eval_interval 20
   

   --root_path [Train data path]

   --test_path [Test data path]

   --eval_interval [Evaluation epoch]

3. Run the below code to test the D-LKA Net on the Synapse dataset.

   python test.py --volume_path ./data/Synapse/ --output_dir './model_out'
   

   --volume_path [Root dir of the test data]

   --output_dir [Directory of your learned weights]

Skin Dataset

Examples are given for the Skin2017 dataset. The other datasets work exactly the same.

Data Preparation

1. Download the dataset from the link above.

2. Prepare the data. Adjust the filespath in the preparation file accordingly.

   cd 2D/skin_code
   python Prepare_ISIC_2017.py
   

   The Data structure should be as follows:

   -ISIC2017
     --/data_train.npy
     --/data_test.npy
     --/data_val.npy
     --/mask_train.npy
     --/mask_test.npy
     --/mask_val.npy
   

Training and Testing

1. Adjust the path in the train_skin_2017.py file for your paths.
2. Run the following line of code:

   python train_skin_2017.py
   

3. For evaluation follow the instruction in the jupyter notebook evaluate_skin.ipynb