DualAttNet: Synergistic Fusion of Image-level and Fine-Grained Disease Attention for Multi-Label Lesion Detection in Chest X-rays

TL;DR

DualAttNet introduces a dual attention mechanism that effectively fuses global and local features for improved multi-label lesion detection in chest X-rays, outperforming existing methods across multiple datasets.

Contribution

The paper proposes a novel dual attention supervised module that enhances lesion detection by combining image-level and fine-grained disease attention, refining feature representations.

Findings

Surpasses baseline models by up to 2.7% mAP and 4.7% AP50.

Effective on multiple datasets including VinDr-CXR, ChestX-ray8, and COVID-19.

Demonstrates improved focus on lesion regions in chest X-ray images.

Abstract

Chest radiographs are the most commonly performed radiological examinations for lesion detection. Recent advances in deep learning have led to encouraging results in various thoracic disease detection tasks. Particularly, the architecture with feature pyramid network performs the ability to recognise targets with different sizes. However, such networks are difficult to focus on lesion regions in chest X-rays due to their high resemblance in vision. In this paper, we propose a dual attention supervised module for multi-label lesion detection in chest radiographs, named DualAttNet. It efficiently fuses global and local lesion classification information based on an image-level attention block and a fine-grained disease attention algorithm. A binary cross entropy loss function is used to calculate the difference between the attention map and ground truth at image level. The generated gradient flow is leveraged to refine pyramid representations and highlight lesion-related features. We evaluate the proposed model on VinDr-CXR, ChestX-ray8 and COVID-19 datasets. The experimental results show that DualAttNet surpasses baselines by 0.6% to 2.7% mAP and 1.4% to 4.7% AP50 with different detection architectures. The code for our work and more technical details can be found at https://github.com/xq141839/DualAttNet.