The Tensorflow code in this repository implements the modifications of the VAE and Ladder VAE frameworks presented in Learning Interpretable Anatomical Features through Deep Generative Models: Application to Cardiac Remodeling and Explainable Shape Analysis through Deep Hierarchical Generative Models: Application to Cardiac Remodelling papers.
Make sure you have Python 3.4 and Tensorflow installed.
The architecture and training details can be configured in the config/config.json file.
To train the network please run:
python training.py --config=jsons/config.json
This implementation was inspired by geosada Tensorflow implementation of the LVAE original paper. If you find this work useful, please cite the following papers:
[1] Biffi, C., et al. Explainable Shape Analysis through Deep Hierarchical Generative Models: Application to Cardiac Remodelling Submitted for review to IEEE Transactions on Medical Imaging.
[2] Biffi, C., et al. Learning Interpretable Anatomical Features through Deep Generative Models: Application to Cardiac Remodeling International Conference on Medical Image Computing and Computer-Assisted Intervention. Springer, Cham, 2018.