Learning Temporally Equivariance for Degenerative Disease Progression in OCT by Predicting Future Representations

TL;DR

This paper introduces a time-equivariant contrastive learning method that captures disease progression in longitudinal OCT scans, outperforming existing methods in predicting AMD progression.

Contribution

The paper proposes a novel time-equivariant contrastive learning approach that models disease progression by predicting future representations from past scans.

Findings

Outperforms existing equivariant contrastive methods in AMD progression prediction

Effectively captures temporal anatomical changes in longitudinal imaging

Improves early detection of disease progression

Abstract

Contrastive pretraining provides robust representations by ensuring their invariance to different image transformations while simultaneously preventing representational collapse. Equivariant contrastive learning, on the other hand, provides representations sensitive to specific image transformations while remaining invariant to others. By introducing equivariance to time-induced transformations, such as disease-related anatomical changes in longitudinal imaging, the model can effectively capture such changes in the representation space. In this work, we propose a Time-equivariant Contrastive Learning (TC) method. First, an encoder embeds two unlabeled scans from different time points of the same patient into the representation space. Next, a temporal equivariance module is trained to predict the representation of a later visit based on the representation from one of the previous visits and the corresponding time interval with a novel regularization loss term while preserving the invariance property to irrelevant image transformations. On a large longitudinal dataset, our model clearly outperforms existing equivariant contrastive methods in predicting progression from intermediate age-related macular degeneration (AMD) to advanced wet-AMD within a specified time-window.