Automatic Whole Heart Segmentation Using a Two-stage U-Net Framework and an Adaptive Threshold Window

If you find this code is useful for your research, please consider citing:

@ARTICLE{8737680, 
author={T. {Liu} and Y. {Tian} and S. {Zhao} and X. {Huang} and Q. {Wang}}, 
journal={IEEE Access}, 
title={Automatic Whole Heart Segmentation Using a Two-Stage U-Net Framework and an Adaptive Threshold Window}, 
year={2019}, 
volume={7}, 
number={}, 
pages={83628-83636}, 
keywords={Heart;Image segmentation;Training;Microsoft Windows;Task analysis;Medical diagnostic imaging;Whole heart segmentation;U-Net;adaptive threshold window;segmentation}, 
doi={10.1109/ACCESS.2019.2923318}, 
ISSN={2169-3536}, 
month={},}

Installation

python 3.5
tensorflow-gpu	1.12.0
NiftyNet	0.4.0

download the data set to ./data

Step one

1. Get train image：python step_one/get_image.py
2. Get label image：python step_one/get_label.py
3. Get weight map：python step_one/heatmap.py
4. Get test image: python step_one/get_test.py
5. Run the configuration file for training: python net_segment.py train -c ini/train_whole_heart.ini
6. Run the configuration for inference：python net_segment.py inference -c ini/inference_whole_heart.ini
7. Display results：python step_one/show_result.py
8. View score：python step_one/score_of_whole_heart.py

Step two

1. Get train image：python step_two/get_image.py
2. Get label image：python step_two/get_label.py
3. Get weight map：python step_two/heatmap,py
4. Get test image: python step_two/get_test.py
5. Run the configuration file for training:

• python net_segment.py train -c ini/train_0.ini
• python net_segment.py train -c ini/train_1.ini
• python net_segment.py train -c ini/train_2.ini
• python net_segment.py train -c ini/train_3.ini
• python net_segment.py train -c ini/train_4.ini
• python net_segment.py train -c ini/train_5.ini
• python net_segment.py train -c ini/train_6.ini

6. Run the configuration for inference：

• python net_segment.py nference -c ini/inference_0.ini
• python net_segment.py inference -c ini/inference_1.ini
• python net_segment.py inference -c ini/inference_2.ini
• python net_segment.py inference -c ini/inference_3.ini
• python net_segment.py inference -c ini/inference_4.ini
• python net_segment.py inference -c ini/inference_5.ini
• python net_segment.py inference -c ini/inference_6.ini

7. Display results：python step_two/show_result.py
8. View score：python step_two/score_of_ever_part.py