# Brain-Tissue-Segment-Keras
Network: FCN, UNet, Unet+Resblock(DeepUNet), Squeeze+UNet, DenseNet+UNet(DenseUNet)...

using small patch_size(32,32)
-------------------------------------------------------------------------------------------
---config.py: Set parameters
Including:PATCH_SIZE,MODEL_TYPE,BATCH_SIZE...


---data_handling_2d_patch.py
Make Training Set and Validation Set
output:2d_patch/

---train_main_2d_patch.py
main function
output: outputs/ (you can see model.h5 and result.csv)
output: logs/ (you can see loss curve in tensorboard)
                  tensorboard --logdir=logs

---predict_test_2d.py
poutput: predict/(you can see the predicted result of Test Set)

-----------------------------------------------------------------------------------------
Some Reference（just reference!!! not absolutely same!）

UNet+ResBlock(DeepUNet)：
https://arxiv.org/pdf/1711.10684.pdf

DenseNet+UNet：
https://arxiv.org/pdf/1611.09326.pdf

Squeeze+UNet
https://arxiv.org/abs/1602.07360