Retinal OCT disease classification with variational autoencoder regularization

Max-Heinrich Laves, Sontje Ihler, Lüder A. Kahrs, Tobias Ortmaier

Code for our abstract paper accepted at 33rd international conference on Computer Assisted Radiology and Surgery (CARS) 2019.

Paper link: https://arxiv.org/pdf/1904.00790.pdf

Abstract

This paper describes a method based on variational autoencoder regularization that improves OCT retinal disease classification performance when using a limited amount of labeled data.

Figure 1: t-SNE projection of the latent space.

BibTeX

@inproceedings{Laves2019CARS,
    author = {Max-Heinrich Laves and Sontje Ihler and L{\"u}der A. Kahrs and Tobias Ortmaier},
    title = {Retinal OCT disease classification with variational autoencoder regularization},
    booktitle={arXiv},
    note = {https://arxiv.org/abs/1904.00790},
    year = {2019}
}

Contact

Max-Heinrich Laves
laves@imes.uni-hannover.de
@MaxLaves

Institute of Mechatronic Systems
Leibniz Universität Hannover
An der Universität 1, 30823 Garbsen, Germany