# Modeling Barrett's Esophagus Progression using Geometric Variational Autoencoders

arXiv: https://arxiv.org/abs/2303.12711 · HTML: https://mlsanity.com/paper/2303.12711/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 11 of its 19 references that are in the map.
- First posted on arXiv 17 Mar 2023.

## Builds on (most cited first, 11 of 11)

- [Group Equivariant Convolutional Networks](https://mlsanity.com/paper/1602.07576/) (2016)
- [Hyperspherical Variational Auto-Encoders](https://mlsanity.com/paper/1804.00891/) (2018)
- [Latent Space Oddity: on the Curvature of Deep Generative Models](https://mlsanity.com/paper/1710.11379/) (2017)
- [Constant Curvature Graph Convolutional Networks](https://mlsanity.com/paper/1911.05076/) (2019)
- [Roto-Translation Covariant Convolutional Networks for Medical Image Analysis](https://mlsanity.com/paper/1804.03393/) (2018)
- [Mixed-curvature Variational Autoencoders](https://mlsanity.com/paper/1911.08411/) (2019)
- [The Riemannian Geometry of Deep Generative Models](https://mlsanity.com/paper/1711.08014/) (2017)
- [Metrics for Deep Generative Models](https://mlsanity.com/paper/1711.01204/) (2017)
- [Metrics for Probabilistic Geometries](https://mlsanity.com/paper/1411.7432/) (2014)
- [Geometry-Aware Hamiltonian Variational Auto-Encoder](https://mlsanity.com/paper/2010.11518/) (2020)
- [Orientation-Disentangled Unsupervised Representation Learning for Computational Pathology](https://mlsanity.com/paper/2008.11673/) (2020)

## Abstract (arXiv)

Early detection of Barrett's Esophagus (BE), the only known precursor to Esophageal adenocarcinoma (EAC), is crucial for effectively preventing and treating esophageal cancer. In this work, we investigate the potential of geometric Variational Autoencoders (VAEs) to learn a meaningful latent representation that captures the progression of BE. We show that hyperspherical VAE (S-VAE) and Kendall Shape VAE show improved classification accuracy, reconstruction loss, and generative capacity. Additionally, we present a novel autoencoder architecture that can generate qualitative images without the need for a variational framework while retaining the benefits of an autoencoder, such as improved stability and reconstruction quality.