BAT-Net

Code for paper titled "Automated segmentation of the human supraclavicular fat depot via deep neural network in water-fat separated magnetic resonance images"

A sketch of the developed deep learning methods:

An overview of the whole network. It consists of three combining 2D U-Net-like networks and a 3D fusion network to mimic the workflow of physicians for characterizing BAT regions and to efficiently encode the multi-modal information and extract the 3D context information from multi-modal MRI scans for the segmentation of the BAT. The three combining 2D networks leverage multi-modal information and comprehensive 2D context information in axial, coronal, and sagittal planes to conduct the preliminary segmentation and the 3D fusion network combines multi-modal information, 3D context information and preliminary segmentation results for obtaining a fine-tuning segmentation.

Requirement:

Python 2.7.3
tensorflow 1.9.0
Keras 2.2.2
keras-contrib 2.0.8
pandas 0.24.2
scikit-image 0.14.0
scikit-learn 0.19.2
SimpleITK 1.1.0

Guideline for utilizing:

Three-combining 2D segmentation network component:

(1) Editor /Three_combining_2D_segmentation_network/config.py file:

Edit config file to assign parameters such as GPU device (A sample config are provided).

(2) Data should be organized as:

Input_directory_path:    
    /data/FF/ # Fat Fraction modality
        /*.nii    
    data/T2S/ # T2* modality
        /*.nii       
    data/F/ # Fat modality
        /*.nii   
    data/W/ # Water modality
        /*.nii   
    data/Labels/ # Manual annotation
        /*.nii   

(3) Training data and test data split can be assigned as follows:

# Training data
Input_directory_path:    
    TrainingData/FF.txt # Fat Fraction modality  
    TrainingData/T2S.txt # T2* modality
    TrainingData/F.txt # Fat modality  
    TrainingData/W.txt # Water modality 
    TrainingData/Label.txt # Manual annotation

# Test data
Input_directory_path:    
    TestData/FF.txt # Fat Fraction modality
    TestData/T2S.txt # T2* modality 
    TestData/F.txt # Fat modality
    TestData/W.txt # Water modality
    TestData/Label.txt # Manual annotation

(4) Data preparing

python /Three_combining_2D_segmentation_network/prepare_data.py 
--data_folder Input_directory_path 
--project_folder project_folder

(4) training the model:

python /Three_combining_2D_segmentation_network/main.py 
--project_folder project_folder 
--mode 'train' 
--learning_rate 0.0001 
--epochs 100 
--batch_size = 30

(5) predicton on the unseen data (Optional, can be utilized when evaluating the performance of the 2D network):

python /Three_combining_2D_segmentation_network/main.py 
--project_folder project_folder 
--mode 'test' 

(6) Post-Processing and evaluation (2D)) (Optional, can be utilized when evaluating the performance of the 2D network):

python post_processing_and_evaluation.py
--data_folder Input_directory_path 
--project_folder project_folder

3D fusion network component:

(1) Editor /ThreeD_fusion_net/config.py file:

Setup the seed and other parameters in /ThreeD_fusion_net/config.py. please save the changes before next step. (A sample config are provided).

(2) Data preprocessing:

python ThreeD_fusion_net/preprocessing.py --input_data_root Input_directory_path

(3) training the model:

python ThreeD_fusion_net/train.py
--learning_rate 0.0001 
--epochs 39 
--batch_size = 8

(4): predicton on the unseen data:

python ThreeD_fusion_net/predict.py --output_folder project_directory_path

(5): performance evaluation

python ThreeD_fusion_net/evaluate_bat.py --output_folder project_directory_path

(6): Postprocessing3D

python ThreeD_fusion_net/Postprocessing3D.py --input_folder project_directory_path --output_folder project_directory_path