Edge-aware Feature Aggregation Network for Polyp Segmentation

TL;DR

This paper tackles automatic polyp segmentation in colonoscopy images under challenging scale variation and blurred boundaries. It introduces EFA-Net, an architecture built on a Res2Net backbone that fuses cross-level and multi-scale features via three modules: Edge-aware Guidance Module (EGM), Scale-aware Convolution Module (SCM), and Cross-level Fusion Module (CFM). By weighting cross-level fused features with learned edge cues and producing multiple side-out maps, EFA-Net achieves state-of-the-art or near-state-of-the-art performance across five datasets, validated by Dice, IoU, and edge-aware metrics, with robust generalization to unseen data. The approach demonstrates strong potential for practical deployment, and the authors provide open-source code and aim to extend efficiency for real-time clinical use.

Abstract

Precise polyp segmentation is vital for the early diagnosis and prevention of colorectal cancer (CRC) in clinical practice. However, due to scale variation and blurry polyp boundaries, it is still a challenging task to achieve satisfactory segmentation performance with different scales and shapes. In this study, we present a novel Edge-aware Feature Aggregation Network (EFA-Net) for polyp segmentation, which can fully make use of cross-level and multi-scale features to enhance the performance of polyp segmentation. Specifically, we first present an Edge-aware Guidance Module (EGM) to combine the low-level features with the high-level features to learn an edge-enhanced feature, which is incorporated into each decoder unit using a layer-by-layer strategy. Besides, a Scale-aware Convolution Module (SCM) is proposed to learn scale-aware features by using dilated convolutions with different ratios, in order to effectively deal with scale variation. Further, a Cross-level Fusion Module (CFM) is proposed to effectively integrate the cross-level features, which can exploit the local and global contextual information. Finally, the outputs of CFMs are adaptively weighted by using the learned edge-aware feature, which are then used to produce multiple side-out segmentation maps. Experimental results on five widely adopted colonoscopy datasets show that our EFA-Net outperforms state-of-the-art polyp segmentation methods in terms of generalization and effectiveness. Our implementation code and segmentation maps will be publicly at https://github.com/taozh2017/EFANet.

Figures (8)

• Figure 1: Illustration of different challenges in polyp segmentation, i.e., 1) The same type of polyps has a diversity of scale (see the first three columns), and 2) the blurred boundary between a polyp and its surrounding mucosa (see the fourth and fifth columns).
• Figure 2: The architecture of the proposed EFA-Net, which consists of edge-aware guidance module, scale-aware convolution module, and cross-level fusion module..
• Figure 3: Architecture of the proposed scale-aware convolution module.
• Figure 4: Architecture of the proposed cross-level fusion module.
• Figure 5: Qualitative results of our model and seven representative methods.