Graph CNNs for population graphs
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information. fixed bugs in creating paths May 3, 2017
LICENSE.txt Initial commit Apr 28, 2017 update copyright Apr 28, 2017
subject_IDs.txt Fix bug for setting depth parameter Jun 12, 2017

Graph CNNs for population graphs: classification of the ABIDE dataset

This code provides a python - Tensorflow implementation of graph convolutional networks (GCNs) for semi-supervised disease prediction using population graphs, as described in:

Parisot, S., Ktena, S. I., Ferrante, E., Lee, M., Moreno, R. G., Glocker, B., & Rueckert, D. (2017).
Spectral Graph Convolutions for Population-based Disease Prediction.
arXiv preprint arXiv:1703.03020.

We provide an implementation applied to the ABIDE dataset for diagnosis of Autism Spectrum Disorder. We also provide the list of scans from the ADNI dataset used in our experiments. Each element of the list is in the format {SUBJECT_ID}_{ACQUISITION_MONTH}


To run the programme, you will need to install the implementation of graph convolutional networks (GCN) by Kipf et al. This project is only compatible with our forked GCN project.

The root folder in (line 12) and (line 17) has to be updated to the folder were the data will be stored.

Next, to install, organise and pre-process the ABIDE database: python


To run the programme with default parameters:


To get a detailed description of parameters:

python --help 


tensorflow (>0.12)


Please cite our paper if you use this code in your own work:

  title={Spectral Graph Convolutions on Population Graphs for Disease Prediction}, 
  author={Parisot, Sarah and Ktena, Sofia Ira and Ferrante, Enzo and Lee, Matthew and Moreno, Ricardo Guerrerro and Glocker, Ben and Rueckert, Daniel}, 
  journal={arXiv preprint arXiv:1703.03020},