Distribution-Aware Replay for Continual MRI Segmentation

TL;DR

This paper introduces a distribution-aware replay method for continual MRI segmentation that mitigates forgetting and detects model failure by auto-encoding features and leveraging learned feature distributions, improving performance without privacy concerns.

Contribution

It proposes a novel distribution-aware replay strategy that combines feature auto-encoding with distribution modeling to enhance continual learning in medical image segmentation.

Findings

Effective in hippocampus and prostate MRI segmentation

Reduces catastrophic forgetting in continual learning

Detects out-of-distribution failures

Abstract

Medical image distributions shift constantly due to changes in patient population and discrepancies in image acquisition. These distribution changes result in performance deterioration; deterioration that continual learning aims to alleviate. However, only adaptation with data rehearsal strategies yields practically desirable performance for medical image segmentation. Such rehearsal violates patient privacy and, as most continual learning approaches, overlooks unexpected changes from out-of-distribution instances. To transcend both of these challenges, we introduce a distribution-aware replay strategy that mitigates forgetting through auto-encoding of features, while simultaneously leveraging the learned distribution of features to detect model failure. We provide empirical corroboration on hippocampus and prostate MRI segmentation.