Few-Shot Domain Adaptation with Polymorphic Transformers

TL;DR

This paper introduces Polyformer, a polymorphic transformer module that enables effective few-shot domain adaptation in medical image segmentation by leveraging prototype embeddings and minimal parameter updates.

Contribution

The paper proposes a novel Polyformer layer that facilitates few-shot domain adaptation by extracting source prototypes and updating only a projection layer, reducing overfitting and improving robustness.

Findings

Polyformer improves segmentation accuracy on unseen domains.

Effective with minimal annotated data in target domain.

Reduces overfitting by freezing most model weights during adaptation.

Abstract

Deep neural networks (DNNs) trained on one set of medical images often experience severe performance drop on unseen test images, due to various domain discrepancy between the training images (source domain) and the test images (target domain), which raises a domain adaptation issue. In clinical settings, it is difficult to collect enough annotated target domain data in a short period. Few-shot domain adaptation, i.e., adapting a trained model with a handful of annotations, is highly practical and useful in this case. In this paper, we propose a Polymorphic Transformer (Polyformer), which can be incorporated into any DNN backbones for few-shot domain adaptation. Specifically, after the polyformer layer is inserted into a model trained on the source domain, it extracts a set of prototype embeddings, which can be viewed as a "basis" of the source-domain features. On the target domain, the polyformer layer adapts by only updating a projection layer which controls the interactions between image features and the prototype embeddings. All other model weights (except BatchNorm parameters) are frozen during adaptation. Thus, the chance of overfitting the annotations is greatly reduced, and the model can perform robustly on the target domain after being trained on a few annotated images. We demonstrate the effectiveness of Polyformer on two medical segmentation tasks (i.e., optic disc/cup segmentation, and polyp segmentation). The source code of Polyformer is released at https://github.com/askerlee/segtran.