Detect-liver-tumor-by-CNN3D-and-UNET3D

1. About this project :

• It is my graduate thesis in university (2019).
• Topic: Detect liver tumors in CT image
• Dataset: Liver tumor Segmentation Challenge (LiTS) contain 131 contrast-enhanced CT images provided by hospital around the world. If you'd like to download it, you can find in this link: https://competitions.codalab.org/competitions/17094
• Used models: CNN3D and UNet3D (3D mean using convolution 3D (3x3x3), max pooling 3D (3x3x3),...)

2. Requirements about enviroment

• I use python==3.7.0 and tensorfow==1.14.0.
• About the library, you can install all packages in requirements.txt by run this command:

pip install -r requirements.txt

3. Analyse dataset

• Open file: Analyse_data.ipynb

• Function Histogram() to draw the data's histogram. See below chart:

• Function Information to log the tumor information (size of tumor in whole image)

• Function PixelRatio to draw the ratio chart between tumor and back ground. See below chart:

3. Training and testing

3.1 Training

3.2 Testing

4. Results

5. References

• A list of papers refer to this project:
  • Detection-aided liver lesion segmentation using deep learning.
  • Automatic Liver Lesion Segmentation Using A Deep Convolutional Neural Network Method.
  • V-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentation
• A list of other great sources that were sources of inspiration:
  • liverseg-2017-nipsws
  • MICCAI-LITS2017