Transfer Risk Map: Mitigating Pixel-level Negative Transfer in Medical Segmentation

TL;DR

This paper introduces a transfer risk map and a weighted fine-tuning method to focus on high-risk regions in medical image segmentation, effectively reducing negative transfer and improving performance across datasets and scenarios.

Contribution

It proposes a transferability-guided risk map and a map-weighted loss for pixel-level negative transfer mitigation in medical segmentation, a novel approach in this domain.

Findings

Achieved 4.37% improvement on FeTS2021 dataset.

Achieved 1.81% improvement on iSeg2019 dataset.

Demonstrated robustness in few-shot learning scenarios.

Abstract

How to mitigate negative transfer in transfer learning is a long-standing and challenging issue, especially in the application of medical image segmentation. Existing methods for reducing negative transfer focus on classification or regression tasks, ignoring the non-uniform negative transfer risk in different image regions. In this work, we propose a simple yet effective weighted fine-tuning method that directs the model's attention towards regions with significant transfer risk for medical semantic segmentation. Specifically, we compute a transferability-guided transfer risk map to quantify the transfer hardness for each pixel and the potential risks of negative transfer. During the fine-tuning phase, we introduce a map-weighted loss function, normalized with image foreground size to counter class imbalance. Extensive experiments on brain segmentation datasets show our method significantly improves the target task performance, with gains of 4.37% on FeTS2021 and 1.81% on iSeg2019, avoiding negative transfer across modalities and tasks. Meanwhile, a 2.9% gain under a few-shot scenario validates the robustness of our approach.