This is the official PyTorch implementation of the paper - "Multi-Aperture Fusion of Transformer-Convolutional Network (MFTC-Net) for 3D Medical Image Segmentation and Visualization".

Multi-Aperture Fusion of Transformer-Convolutional Network (MFTC-Net) for 3D Medical Image Segmentation and Visualization

Set Up Set Up Environment

python3.10 -m venv MFTCNet_env 
source MFTCNet_env/bin/activate 
pip install -r requirements.txt

Dataset

Download data from: https://www.synapse.org/#!Synapse:syn3193805/wiki/89480

Structure of data folders

data/
 |---imagesTr/
     |---img0001.nii.gz
     |---img0002.nii.gz
 |---labelsTr/
     |---label0001.nii.gz
     |---label0002.nii.gz
 |---dataset.json

Running the Code

This repository is built upon the foundational work provided in Synapse.

Training

Before training the configs.json file should be filled:

Hyperparameters Configuration

The following settings can be adjusted in the config.py to configure the model training and data management:

• data_dir: Set the directory path for dataset storage.
• saved_model_dir: Set the directory path where trained models and checkpoints will be saved.
• num_samples: Define the number of samples used in training process.
• num_classes: Specify the number of target classes in the dataset + background.
• input_size: Set the size of the input images or data.
• input_channels: Define the number of input channels for the data (e.g., grayscale=1, RGB=3).
• feature_size: Set the dimension of the feature vectors extracted by the model.
• use_checkpoint: Enable or disable the use of model checkpoints for training continuation.
• learning_rate: Set the initial learning rate for the optimizer.
• weight_decay: Define the weight decay (L2 penalty) rate for regularization.
• max_iterations: Specify the maximum number of iterations (or training steps).
• eval_num: Set the frequency of evaluations (every eval_num iterations) during training.

train the model

python3.10 main.py  

Results

Quantitative

Visulization

Citation

If any part of this code is used, please give appropriate citations to our paper.

Contact

If you have any questions, please email sshabani@unr.edu to discuss with the authors.