Learning Temporally Equivariance for Degenerative Disease Progression in OCT by Predicting Future Representations
This work addresses the challenge of capturing disease-related anatomical changes over time in medical imaging for applications like AMD progression prediction, representing an incremental improvement over existing equivariant contrastive methods.
The paper tackled the problem of predicting disease progression in longitudinal OCT imaging by introducing a time-equivariant contrastive learning method, which outperformed existing equivariant methods in predicting progression from intermediate to advanced AMD within a time-window.
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.