Lung nodule segmentation

• Lung nodule segmentation implemented in Pytorch.
• Read and implement 2 papers:
  • ResUNet++: An Advanced Architecture for Medical Image Segmentation
  • Dual Attention Network for Scene Segmentation

Requirements

• Python >= 3.5 (3.6 recommended).

  pip install -r requirements.txt
  

Dataset

• Luna16 part 1
• Luna16 part 2

Folder Structure

.
├── backbone
│   ├── __init__.py 
│   ├── modules.py
│   ├── res_unet_plus.py
│   ├── res_unet.py
│   └── unet.py
├── checkpoints
│   └── default
│       └── exp8_512.pt - pretrained model with image size 512 x 512
├── config
│   └── default.yaml - holds configuration for training
├── dataset
│   └── dataloader.py - anything about data loading goes here
├── inference.py
├── luna_mask_extraction.py - extracts raw data to image, mask
├── lung_segment_png.py
├── preprocess.py
├── README.md
├── requirements.txt
├── train.py
└── utils
    ├── augmentation.py
    ├── hparams.py
    ├── __init__.py
    ├── logger.py
    └── metrics.py

6 directories, 20 files

Usage

• Training
  • From scratch:

    python train.py --name "default" --config "config/default.yaml"
    

  • Load from:

    python train.py --name "default" --config "config/default.yaml" \ 
    --load_from "checkpoints/default/exp8_512.pt"
    

• Inference:

python inference.py

Results