TINC: Temporally Informed Non-Contrastive Learning for Disease Progression Modeling in Retinal OCT Volumes
This work addresses the problem of noisy annotations and class imbalance in medical imaging for clinicians, by improving disease progression prediction in age-related macular degeneration, though it is incremental as it builds on non-contrastive learning methods.
The paper tackled disease progression modeling in retinal OCT volumes by developing a temporally informed non-contrastive learning method (TINC) that leverages temporal information from longitudinal data without heavy augmentations or negative pairs, resulting in outperforming existing models in predicting the risk of conversion from intermediate to late wet-AMD within a time frame.
Recent contrastive learning methods achieved state-of-the-art in low label regimes. However, the training requires large batch sizes and heavy augmentations to create multiple views of an image. With non-contrastive methods, the negatives are implicitly incorporated in the loss, allowing different images and modalities as pairs. Although the meta-information (i.e., age, sex) in medical imaging is abundant, the annotations are noisy and prone to class imbalance. In this work, we exploited already existing temporal information (different visits from a patient) in a longitudinal optical coherence tomography (OCT) dataset using temporally informed non-contrastive loss (TINC) without increasing complexity and need for negative pairs. Moreover, our novel pair-forming scheme can avoid heavy augmentations and implicitly incorporates the temporal information in the pairs. Finally, these representations learned from the pretraining are more successful in predicting disease progression where the temporal information is crucial for the downstream task. More specifically, our model outperforms existing models in predicting the risk of conversion within a time frame from intermediate age-related macular degeneration (AMD) to the late wet-AMD stage.