Deep-ADMM-Net-master
所属分类：其他
开发工具：matlab
文件大小：88459KB
下载次数：8
上传日期：2019-04-27 20:45:41
上 传 者：
林一亦
说明： 基于Deep-ADMM-Net的CT重建算法
(CT reconstruction algorithm based on Deep-ADMM-Net)
文件列表:
Deep-ADMM-Net-master\config.m (475, 2017-04-25)
Deep-ADMM-Net-master\data\Brain_data\Brain_data1.mat (376122, 2017-04-25)
Deep-ADMM-Net-master\data\Brain_data\Brain_data2.mat (498584, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-01.mat (94331, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-02.mat (86540, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-03.mat (104789, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-04.mat (85597, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-05.mat (96970, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-06.mat (100607, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-07.mat (108057, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-08.mat (106140, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-09.mat (117854, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-10.mat (74679, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-100.mat (135014, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-11.mat (163242, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-12.mat (143600, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-13.mat (131237, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-14.mat (112500, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-15.mat (99133, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-16.mat (98307, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-17.mat (106314, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-18.mat (82722, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-19.mat (93774, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-20.mat (103252, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-21.mat (96922, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-22.mat (108829, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-23.mat (70440, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-24.mat (131130, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-25.mat (86703, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-26.mat (106207, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-27.mat (136561, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-28.mat (118289, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-29.mat (101033, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-30.mat (125706, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-31.mat (106771, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-32.mat (124899, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-33.mat (112519, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-34.mat (128363, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-35.mat (104665, 2017-04-25)
Deep-ADMM-Net-master\data\ChestTrain\im-36.mat (157430, 2017-04-25)
... ...
# Deep-ADMM-Net
***********************************************************************************************************
This is a testing and training code for Deep ADMM-Net in "Deep ADMM-Net for Compressive Sensing MRI" (NIPS 2016)
If you use this code, please cite our paper:
[[1] Yan Yang, Jian Sun, Huibin Li, Zongben Xu. Deep ADMM-Net for Compressive Sensing MRI, NIPS(2016).](http://gr.xjtu.edu.cn/web/jiansun/publications])
http://gr.xjtu.edu.cn/web/jiansun/publications
All rights are reserved by the authors.
Yan Yang -2017/04/05. For more detail, feel free to contact: yangyan92@stu.xjtu.edu.cn
***********************************************************************************************************
## Usage:
1. For testing the trained network
1). Load trained network with different stages in main_ADMM_Net_test.m.
If you apply ADMM-Net to reconstruct other MR images, it is best to re-train the models.
The models in './net/network_20' are trained from 100 real MR trainging images with 20% sampling rate.
The models in './net/network_30' are trained from 100 real MR trainging images with 30% sampling rate.
2). Load sampling pattern with different sampling ratios in main_ADMM_Net_test.m
The mask in './mask/mask_20' is a pseudo radial sampling pattern with 20% sampling rate.
3). Load test image in main_ADMM_Net_test.m
The images in './data/Brain_data' are real-valued brain MR images.
The images in './data/Chest_data' are 50 real-valued chest MR testing images in our paper.
4). Network setting is in 'config.m '.
5). To test our ADMM-Net, run 'main_ADMM_Net_test.m'
2. For training the networks
1). The training chest dataset is in './data/ChestTrain_20'.
Run 'Gen_traindata.m' to generate training data, and load corresponding sampling pattern in this operation.
2). Modify the network setting and trainging setting in 'config.m '.
3). To train ADMM-Net by L-BFGS algorithm, run 'main_netTrain.m' .
4). After training, the trained network and the training error are saved in './Train_output'.
***********************************************************************************************************
The testing result of the demo images.
1) Brain_data1.(20% sampling rate)
|--------------| re_LOss | re_PSnr |
|--------------|-----------|-----------|
| net-stage7- | 0.0578 | 35.60 |
| net-stage14 | 0.0562 | 35.83 |
| net-stage15 | 0.0561 | 35.85 |
2) Brain_data2.(20% sampling rate)
|--------------| re_LOss | re_PSnr |
|--------------|-----------|-----------|
| net-stage7- | 0.0957 | 30.40 |
| net-stage14 | 0.0929 | 30.65 |
| net-stage15 | 0.0927 | 30.67 |
近期下载者：
相关文件：
收藏者：