The goal of the project is to segment prostate cancer tissues by gleason scoring system.
Currently, we're using deep learning model - U-net architecture.
- extract_model/ : extract weight parameters from pre-trained U-net caffe model to npy's
- param/ : training Gleason_U-net weight & bias parameters
- testimages/ : sample test images
- unet.py : Our Gleason_U-net object
- coco.py : training code on cocodatasets
- test.py : test out Gleason_Unet and plot
- resnet.py : Resnet-152 classifier for gleason scoring
- init_resnet.py : initializer of resnet weight parameters
- python3
- tensorflow
- numpy
- scikit-image
- matplotlib
- pycocotools (optional)