Few-Shot Medical Image Segmentation via a Region-enhanced Prototypical Transformer

TL;DR

This paper presents a novel few-shot medical image segmentation method called Region-enhanced Prototypical Transformer (RPT), which uses regional prototypes and a bias-alleviating transformer to improve segmentation accuracy amid limited data and high intra-class variability.

Contribution

The paper introduces a region subdivision strategy and a self-selection mechanism within a transformer framework to enhance prototype quality for few-shot medical image segmentation.

Findings

RPT outperforms state-of-the-art FSMS methods on three datasets.

Regional prototypes improve segmentation accuracy.

Iterative optimization refines global prototypes effectively.

Abstract

Automated segmentation of large volumes of medical images is often plagued by the limited availability of fully annotated data and the diversity of organ surface properties resulting from the use of different acquisition protocols for different patients. In this paper, we introduce a more promising few-shot learning-based method named Region-enhanced Prototypical Transformer (RPT) to mitigate the effects of large intra-class diversity/bias. First, a subdivision strategy is introduced to produce a collection of regional prototypes from the foreground of the support prototype. Second, a self-selection mechanism is proposed to incorporate into the Bias-alleviated Transformer (BaT) block to suppress or remove interferences present in the query prototype and regional support prototypes. By stacking BaT blocks, the proposed RPT can iteratively optimize the generated regional prototypes and finally produce rectified and more accurate global prototypes for Few-Shot Medical Image Segmentation (FSMS). Extensive experiments are conducted on three publicly available medical image datasets, and the obtained results show consistent improvements compared to state-of-the-art FSMS methods. The source code is available at: https://github.com/YazhouZhu19/RPT.