ISBI2018_PETCT_Segmentation

This repository contains the code (in TensorFlow) for "3D fully convolutional networks for co-segmentation of tumors on PET-CT images" paper (ISBI 2018). Compared to the previous semi-automated methods, this method is highly automated without manually user-defined seeds. And one extended journal version with much more details is under revision.

CT/PET Segmentation Results on One Patient

1. CT image

2. PET_SUV image

3. Ground Truth Segmentation on CT image

4. Ground Truth Segmentation on PET_SUV image

5. Prediction on CT image

6. Prediction on PET_SUV image

7. Wrong Predictions on CT image

8. Wrong Predictions on PET_SUV image

Dependencies

• Python2.7
• TensorFlow(1.4.0+)
• DLTK
• other libraries

Citation

If you find this useful, please cite our work as follows:

@INPROCEEDINGS{zszhong2018isbi_petct,
  author={Z. Zhong and Y. Kim and L. Zhou and K. Plichta and B. Allen and J. Buatti and X. Wu},
  booktitle={2018 IEEE 15th International Symposium on Biomedical Imaging (ISBI 2018)},
  title={3D fully convolutional networks for co-segmentation of tumors on PET-CT images},
  year={2018},
  volume={},
  number={},
  pages={228-231},
  keywords={Biomedical imaging;Computed tomography;Image segmentation;Lung;Three-dimensional displays;Tumors;co-segmentation;deep learning;fully convolutional networks;image segmentation;lung tumor segmentation},
  doi={10.1109/ISBI.2018.8363561},
  ISSN={},
  month={April},
}

Contacts

[email protected]

Any discussions or concerns are welcomed!