Code the paper titled Region-of-interest guided Supervoxel Inpainting for Self-supervision published at MICCAI 2020 (https://arxiv.org/pdf/2006.15186.pdf) by Subhradeep Kayal, et al.

Citation

If you find this paper useful for your research, please consider citing the paper:

@inproceedings{kayal2020region,
  title={Region-of-interest guided Supervoxel Inpainting for Self-supervision},
  author={Kayal, Subhradeep and Chen, Shuai and de Bruijne, Marleen},
  booktitle={International Conference on Medical Image Computing and Computer-Assisted Intervention},
  pages={500--509},
  year={2020},
  organization={Springer}
}

Requirements

1. nibabel
2. keras
3. tensorflow
4. numpy
5. sklearn
6. skimage
7. tqdm
8. cv2

Steps (to repeat experiments for BraTS 2018)

1. Get BraTS data from https://www.med.upenn.edu/sbia/brats2018/data.html
2. Create a folder within inpainting called data and put the BraTS data in there, with the folder name MICCAI_BraTS_2018_Data_Training. Make sure there is a folder HGG in there with the relevant files. We only use the images concerning high grade glioma (HGG) in these experiments.
3. Run train.sh
4. Run evaluate.sh when the above is finished

Poster: