FLARE21_Segmentation_DL

My Paper Title

This repository is the official implementation of "Efficient Segmentation of Abdominal Organs using Skip Residual Block UNet Model"

Environments and Requirements

• Windows • CPU, RAM, GPU information • CUDA version (11.3) • python version (3.7)

To install requirements:

pip install -r requirements.txt

Dataset

Please you can donwload the dataset from the following website https://flare.grand-challenge.org/Data/

• The dataset is divided into training and validation. Inside the training and validation further two folders are

created with name images and masks

Preprocessing

• intensity normalization used to process the training, validation and testing images.

Running the data preprocessing code:

python Data_Preprocessing_Flare2021.py

Training

Please run this python code for training:

python Training_Flare_model.py

Trained Models

The trained weights can be download here:

Here is the docker link for retrained on 1k abdominal CT images

https://www.dropbox.com/s/j6ue6it0nrwadjc/aq_enib_flare_seg.tar.gz?dl=0

Inference

To infer the testing cases, run this command:

python prediction_flare21.py

Results

Please check the validation results on leaderboard https://flare.grand-challenge.org/evaluation/challenge/leaderboard/

Acknowledgement

We thank the contributors of public datasets.

If you have any question, please let me know at: [email protected]