• KiTS19 - Kidney Tumor Segmentation Challenge 2019
• • Requirements
  • Getting Started
  • • 1. Download kits19 Dataset
    • 2. Conversion data
    • 3. Train ResUNet for Coarse Kidney Segmentation
    • 4. Capture Coarse Kidney ROI
    • 5. Train DenseUNet for Kidney Tumor Segmentation
    • 6. Evaluation Test Case
    • 7. Post-processing
  • Leaderboard
  • TODO

KiTS19 - Kidney Tumor Segmentation Challenge 2019

KiTS19 is part of the MICCAI 2019 Challenge.
The goal of this challenge is to accelerate the development of reliable kidney and kidney tumor semantic segmentation methodologies.

Requirements

• Python >= 3.6
• PyTorch >= 1.0.0
• pip install -r requirements.txt

Getting Started

1. Download kits19 Dataset

Make sure to install git-lfs before cloning!
Clone kits19 repository (~54 GB)

git clone https://github.com/neheller/kits19.git

2. Conversion data

Conversion nii.gz to npy for easy to read slice (~140 GB)

python conversion_data.py -d "kits19/data" -o "data"

3. Train ResUNet for Coarse Kidney Segmentation

python train_res_unet.py -e 100 -b 32 -l 0.0001 -g 4 -s 512 512 -d "data" --log "runs/ResUNet" --eval_intvl 5 --cp_intvl 5 --vis_intvl 0 --num_workers 8

4. Capture Coarse Kidney ROI

python get_roi.py -b 32 -g 4 -s 512 512 --org_data "kits19/data" --data "data" -r "runs/ResUNet/checkpoint/best.pth" -o "data/roi.json"

5. Train DenseUNet for Kidney Tumor Segmentation

python train_dense_unet.py -e 100 -b 32 -l 0.0001 -g 4 -s 512 512 -d "data" --log "runs/DenseUNet" --eval_intvl 5 --cp_intvl 5 --vis_intvl 0 --num_workers 8

6. Evaluation Test Case

python eval_dense_unet.py -b 32 -g 4 -s 512 512 -d "data" -r "runs/DenseUNet/checkpoint/best.pth" --vis_intvl 0 --num_workers 8 -o "out"

7. Post-processing

python post_processing.py -d "out" -o "out_proc"

Leaderboard

We are the 21st of total 106 teams.

TODO

• Refactor code
• Describe method
• Show result
• Write argument help