dfuc-2022

Diabetic Foot Ulcer Segmentation

This work is a part of submission made to DFUC 2022. DFUC 2022 is hosted by MICCAI 2022, the 25th International Conference on Medical Image Computing and Computer Assisted Intervention.

Usage

Note: Anybody can freely use this work under its license terms. However, the authors shall not be held accountable for the reliability and accuracy of the results and the consequences they may lead to.

Requirements

• python3(This work has been developed using python 3.8.10)

Setup

For setup and basic usage get this repository code in your local device and follow the instructions given.

All the commands given are for linux system. For other system, please search for the equivalent commands accordingly.

Create a virtual environment

python -m venv .venv

Note: This command assumes that you have python3 as your default python version in your device.

Activate the environment

source .venv/bin/activate

Install the required dependencies

pip install -r requirements.txt

Start training

For training, update the train_config file and run the command for training. Parameters in the train_config file can be updated by refering the documentation of training config at docs/train_config_docs.md.

python train.py

For training separate models by making diffrent folds of the training dataset, randomly shuffle the dataset and add the shuffled list to k_fold_utils/k_fold-splits.csv(this csv file needs to be created), then update the train_config_k_fold file and run the command for training. Parameters in the train_config_k_fold file can be updated by refering the documentation of training config at docs/train_config_docs.md.

python train_k_fold.py

Training logs can be visualized using tensorboard. It can be run using following command:

tensorboard --logdir logs/tensorboard/ --port PORT

Replace PORT with port number on which tensorboard is to be run. Visualizations can be obtained in web browser.

Evaluation

For evaluation, update the eval_config file and run the command for evaluation. Parameters in the eval_config file can be updated by refering the documentation of evaluation config at docs/eval_config_docs.md.

python eval.py

For evaluation of models trained by making k folds of the training dataset, update the eval_config_k_fold file and run the command for evaluation. Parameters in the eval_config_k_fold file can be updated by refering the documentation of evaluation config at docs/eval_config_docs.md.

python eval_k_fold.py

License

This work is available under MIT License. See LICENSE for the full license text.