This is an example of the prostate in transversal T2-weighted MR images Segment from MICCAI Grand Challenge:Prostate MR Image Segmentation 2012

The following dependencies are needed:

• numpy >= 1.11.1
• SimpleITK >=1.0.1
• opencv-python >=3.3.0
• tensorflow-gpu ==1.8.0
• pandas >=0.20.1
• scikit-learn >= 0.17.1

1、download trained data,download dataset：https://promise12.grand-challenge.org/download/

2、the file of PROMISE2012Image.csv,is like this format: D:\Data\PROMISE2012\Augmentation\Image/0_1.bmp D:\Data\PROMISE2012\Augmentation\Image/0_10.bmp D:\Data\PROMISE2012\Augmentation\Image/0_2.bmp ...... if you Augmentation trained data path is not D:\Data\PROMISE2012,you should change the csv file path just like this:using C:\Data\ replace D:\Data\PROMISE2012.

3、when data is prepared,just run the vnet_train_predict.py

4、training the model on the GTX1080,it take 20 hours,and i also attach the trained model in the project,you also just use the vnet_train_predict.py file to predict,and get the segmentation result.

5、download trained model:https://pan.baidu.com/s/19E9q6HIUeRB8jpuNhvE2Zg, passworld:obwu

the Challenge result

the loss and model result，the example

• https://github.com/junqiangchen
• email: [email protected]
• WeChat Public number: 最新医学影像技术