CNN patch classifier for segmentation of liver tumors

Workflow:

1. Combine data using combine_data.py script. Data should be organized in 3 sub-folders:
   • ct_scans
   • liver_seg
   • tumors
2. Generate patches h5 file using the generate_patches.py script
3. Train model using training_script.py
4. Predict on test set ct scans using prediction_script.py
5. Analyze results using analyze_script.py

Next Phases (Not in Github):

Run Chan-Vese segmentation with prediction probability map as input 1. Set ct_dir_path to be the path to the ct scans. 2. Set liver_dir_path 3. Set cnn_res_dir_path Then run the 2 Chan-Vese scripts: 1. Chan-Vese step 1 2. Chan-Vese step 2