A Deep Learning method to segment punctate white matter lesions (PWMLs); Brain tumor segmentation.
By: Yalong Liu1, Jie Li1, Ying Wang1, Miaomiao Wang2, Xianjun Li2, Zhicheng Jiao3, Jian Yang2, Xingbo Gao1
- Lab of Video and Image Processing Systems, School of Electronic Engineering, Xidian University, Xi’an 710071, China
- Department of Radiology, The First Affiliated Hospital of Xi'an Jiaotong University, Xi’an 710061, China
- University of North Carolina at Chapel Hill, Chapel Hill, NC 27599, USA
This repository includes:
1.T1WI of 10 patients for test(Full dataset is not allowed to be made public)
2.Full code for model training and inference
3.The link of pre-trained weights on google drive
Python 3.6.3
Tensorflow-gpu 1.12.0
CUDA 9.0
1.Download the repositories and weights.
2.Choose a mode in the main.py('inference' or 'training').
3.Change parameters in configs.py according to the comment in the file.
Enjoy!
This repo borrows tons of code from
matterport/Mask_RCNN
Performance:
Index | --------Original MRI-------- | ---------SOTA---------- | -----Mask R-CNN------ | -----Our Method-------- |
---|
8/77 | |
---|---|
79/67 | |
82/67 | |
83/48 | |
83/54 | |
83/80 |
If you use Refined Segmentation R-CNN in your research, please cite the paper (http://arxiv.org/abs/1906.09684):
@article{Liu2019,
title={Refined Segmentation R-CNN: A Two-stage Convolutional Neural Network for Punctate White Matter Lesion Segmentation in Preterm Infants},
author={Yalong Liu, Jie Li, Ying Wang, Miaomiao Wang, Xianjun Li, Zhicheng Jiao, Jian Yang, Xingbo Gao},
journal={arXiv preprint arXiv:1906.09684},
year={2019}
}