The implementation of PRIOR-Net and PRIOR
The training and test data can be downloaded at https://drive.google.com/drive/folders/1_1Ll5frH72_HwdVmGERrNfnbSpqZqphj. You will get the train.zip, test.zip and PriorNet. First, put the "PriorNet" into the current folder "PRIOR-Net". Next, unzip the train.zip and test.zip to get the folders "train" and "test" and put them in the current directory.
-
Enter the folder "PRIOR-Net" and run "python TFRecordOp.py" and the files in the "TFRecordsFile" will be saved as tfrecord files
-
run "python main_PriorNet.py" and the model will be saved every epoch
- run "python Test_PriorNet.py" to test the files
- Enter the folder "PRIOR" and run "python main_PRIOR.py". Before run the PRIOR, you should first train the PRIOR-Net or directly use the trained model we provided in the folder "PriorNet"
cuda 10.0
python 3.6.13
TensorFlow 1.15.4
Numpy 1.16.0
Scipy 1.2.1
tigre (https://github.com/CERN/TIGRE)
Due to the limitation, we only provide one patient to train and one patient to test. You can use your own datasets to train and test.