PraNet-V2: Dual-Supervised Reverse Attention for Medical Image Segmentation

2025-04-15

AI Summary

To address inaccurate lesion localization, blurry boundaries, and weak background modeling in PraNet-V1 for multi-class medical image segmentation, this paper proposes a Dual-Supervised Reverse Attention (DSRA) mechanism. DSRA introduces explicit background supervision and independent background modeling for the first time, integrated with semantic-enhanced multi-level attention fusion and iterative refinement. Unlike conventional single-class reverse attention, DSRA enables end-to-end multi-class segmentation. Implemented on the Jittor framework, it incorporates multi-level supervised losses and semantic feature fusion. Evaluated on four polyp segmentation datasets, DSRA achieves state-of-the-art performance. When embedded into three mainstream segmentation architectures, it yields up to a 1.36% average Dice score improvement, significantly enhancing lesion localization accuracy and boundary delineation capability.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Multi-instance/Multi-view Learning
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs

Abstract

Accurate medical image segmentation is essential for effective diagnosis and treatment. Previously, PraNet-V1 was proposed to enhance polyp segmentation by introducing a reverse attention (RA) module that utilizes background information. However, PraNet-V1 struggles with multi-class segmentation tasks. To address this limitation, we propose PraNet-V2, which, compared to PraNet-V1, effectively performs a broader range of tasks including multi-class segmentation. At the core of PraNet-V2 is the Dual-Supervised Reverse Attention (DSRA) module, which incorporates explicit background supervision, independent background modeling, and semantically enriched attention fusion. Our PraNet-V2 framework demonstrates strong performance on four polyp segmentation datasets. Additionally, by integrating DSRA to iteratively enhance foreground segmentation results in three state-of-the-art semantic segmentation models, we achieve up to a 1.36% improvement in mean Dice score. Code is available at: https://github.com/ai4colonoscopy/PraNet-V2/tree/main/binary_seg/jittor.

Problem

• Improves multi-class medical image segmentation accuracy
• Enhances polyp segmentation using dual-supervised reverse attention
• Boosts performance in semantic segmentation models via DSRA

Innovation

• Dual-Supervised Reverse Attention module
• Explicit background supervision integration
• Iterative foreground segmentation enhancement