Two-level Semi-supervised Collaborative Medical Image Segmentation with Bidirectional Knowledge Exchange

doi:10.1016/j.media.2025.103853

Two-level Semi-supervised Collaborative Medical Image Segmentation with Bidirectional Knowledge Exchange

2025-10-28

0

PRE

AI

H

Haiyan Wang

陶磊 (Tao Lei)

DOI：10.1016/j.media.2025.103853

Original

Original request for help

Save

Abstract

En 中文

• Proposes a two-level co-training structure where second-level models use ensemble pseudo-labels from the first level to improve segmentation. • Introduces a bidirectional knowledge exchange strategy, where features from second-level models are fed back to first-level models, forming a feedback loop that boosts performance across both levels. • Extensive experiments show that the proposed method outperforms state-of-the-art approaches on multiple benchmarks.

Two-level Semi-supervised Collaborative Medical Image Segmentation with Bidirectional Knowledge Exchange

Abstract

Journal

Organization