The repository contains the core codes of "AtrialJSQnet: A New Framework for Joint Segmentation and Quantification of Left Atrium and Scars Incorporating Spatial and Shape Information". The resposutory includes four folds:
This fold includes some tools written using C++ for the pre-processing of LGE MRI.
This fold includes the python code to train and test the LearnGC, which was published in MedIA2019. In this manuscript, we employed LearnGC for comparison. Note that the scripts to generate the multi-scale patches and the pre-processing code for LearnGC are not included here. For the complete version of this code, please kindly refer to https://github.com/Marie0909/LearnGC.
This fold includes some pre-processing scripts employed in AtrialJSQnet, and some of these scripts aimed to use the generated C++ tools mentioned in the C++ script fold.
This fold includes the python code for training and test the AtrialJSQnet.
The dataset employed in this work is from MICCAI 2018: Atrial Segmentation Challenge.
You may also be interested in following papers:
- Atrial scar quantification via multi-scale CNN in the graph-cuts framework
- Medical Image Analysis on Left Atrial LGE MRI for Atrial Fibrillation Studies: A Review
- AtrialGeneral: Domain Generalization for Left Atrial Segmentation of Multi-center LGE MRIs
For evaluation, you could run LAScarQS2022_evaluate.py. Before runing LAScarQS2022_evaluate.py, you need to install SimpleITK, medpy and hausdorff by running "pip install SimpleITK/medpy/hausdorff". Also, note that this evaluation tool can only work in windows system as we only compiled the c++ tools in windows now, which are saved in the fold namely "tools".
If this code is useful for you, please kindly cite this work via:
@article{journal/MedIA/li2022,
title={Atrial{JSQ}net: a new framework for joint segmentation and quantification of left atrium and scars incorporating spatial and shape information},
author={Li, Lei and Zimmer, Veronika A and Schnabel, Julia A and Zhuang, Xiahai},
journal={Medical Image Analysis},
volume={76},
pages={102303},
year={2022},
publisher={Elsevier}
}
If you have any questions, you are always welcome to contact with [email protected].