Adaptive Bidirectional Displacement for Semi-Supervised Medical Image Segmentation

TL;DR

This paper introduces an Adaptive Bidirectional Displacement method that enhances semi-supervised medical image segmentation by effectively utilizing multiple perturbations and confidence-based sample generation, leading to state-of-the-art results.

Contribution

The novel ABD approach employs bidirectional displacement based on confidence to improve consistency learning in semi-supervised segmentation.

Findings

Achieves state-of-the-art performance on SSMIS tasks.

Effectively suppresses uncontrollable regions in unlabeled data.

Significantly improves baseline methods.

Abstract

Consistency learning is a central strategy to tackle unlabeled data in semi-supervised medical image segmentation (SSMIS), which enforces the model to produce consistent predictions under the perturbation. However, most current approaches solely focus on utilizing a specific single perturbation, which can only cope with limited cases, while employing multiple perturbations simultaneously is hard to guarantee the quality of consistency learning. In this paper, we propose an Adaptive Bidirectional Displacement (ABD) approach to solve the above challenge. Specifically, we first design a bidirectional patch displacement based on reliable prediction confidence for unlabeled data to generate new samples, which can effectively suppress uncontrollable regions and still retain the influence of input perturbations. Meanwhile, to enforce the model to learn the potentially uncontrollable content, a bidirectional displacement operation with inverse confidence is proposed for the labeled images, which generates samples with more unreliable information to facilitate model learning. Extensive experiments show that ABD achieves new state-of-the-art performances for SSMIS, significantly improving different baselines. Source code is available at https://github.com/chy-upc/ABD.