Skip to content

Latest commit

 

History

History
52 lines (31 loc) · 2.45 KB

README.md

File metadata and controls

52 lines (31 loc) · 2.45 KB
Raw

Interpretable Representation Learning of Cardiac MRI via Attribute Regularization

Maxime Di Folco, Cosimin I. Bercea, Emily Chan, Julia A. Schnabel

Accepted at MICCAI 2024

Abstract: Interpretability is essential in medical imaging to ensure that clinicians can comprehend and trust artificial intelligence models. Several approaches have been recently considered to encode attributes in the latent space to enhance its interpretability. Notably, attribute regularization aims to encode a set of attributes along the dimensions of a latent representation. However, this approach is based on Variational AutoEncoder and suffers from blurry reconstruction. In this paper, we propose an Attributed-regularized Soft Introspective Variational Autoencoder that combines attribute regularization of the latent space within the framework of an adversarially trained variational autoencoder. We demonstrate on short-axis cardiac Magnetic Resonance images of the UK Biobank the ability of the proposed method to address blurry reconstruction issues of variational autoencoder methods while preserving the latent space interpretability.

Illustration of the framework

Citation

If you use this code, please cite our paper:

@article{di_folco2024interpretable,
  title={Interpretable Representation Learning of Cardiac MRI via Attribute Regularization},
  author={Di Folco, Maxime and Bercea, Cosmin I and Chan E and Schnabel, Julia A},
  journal={arXiv preprint arXiv:2406.08282},
  year={2024}
}

Contents of this repository:

  • projects/interp_rep/data_prepocessing: notebook to adapt the ACDC dataset to the pipeline and the attributes use to regularize 🚧🚧 This part is still work in progress 🚧🚧

  • projects/interp_rep/: training and testing the proposed method AR-SIVAE and the method they compare against. Based on: IML-CompAI Framework

All computations were performed using Python 3.8.18 and PyTorch 1.12.1

Setup:

For setting up wandb please refer to IML-CompAI Framework.

Overview Deep Learning Framework

Framework overview

Train a network

After applying all of the preprocessing steps, you can run:

python core/Main.py --config_path projects/interp_rep/config/config_***.yaml