CXR-RefineDet: Single-Shot Refinement Neural Network for Chest X-Ray Radiograph Based on Multiple Lesions Detection.

Cong Lin,Yongbin Zheng,Jialun Lin,Xiuchun Xiao

doi:10.1155/2022/4182191

Abstract

The workload of radiologists has dramatically increased in the context of the COVID-19 pandemic, causing misdiagnosis and missed diagnosis of diseases. The use of artificial intelligence technology can assist doctors in locating and identifying lesions in medical images. In order to improve the accuracy of disease diagnosis in medical imaging, we propose a lung disease detection neural network that is superior to the current mainstream object detection model in this paper. By combining the advantages of RepVGG block and Resblock in information fusion and information extraction, we design a backbone RRNet with few parameters and strong feature extraction capabilities. After that, we propose a structure called Information Reuse, which can solve the problem of low utilization of the original network output features by connecting the normalized features back to the network. Combining the network of RRNet and the improved RefineDet, we propose the overall network which was called CXR-RefineDet. Through a large number of experiments on the largest public lung chest radiograph detection dataset VinDr-CXR, it is found that the detection accuracy and inference speed of CXR-RefineDet have reached 0.1686 mAP and 6.8 fps, respectively, which is better than the two-stage object detection algorithm using a strong backbone like ResNet-50 and ResNet-101. In addition, the fast reasoning speed of CXR-RefineDet also provides the possibility for the actual implementation of the computer-aided diagnosis system.

Highlights

Chest X-ray (CXR) is an effective and widely used imaging technique in the diagnosis and screening of lung-related diseases. e imaging principle and structure of chest radiographs are complex, which requires professional radiologists to spend a lot of time to observe carefully
(2) We propose the Information Reuse structure, which solves the problem of low utilization of the original network output features by linking the normalized features back to the network. (3) e proposed object detection model CXR-RefineDet has a good performance between accuracy and speed
In order to solve the problem of weak feature extraction capability of the RefineDet backbone network and low feature utilization of the output feature layer, a high-precision and fast detection speed lung lesion detection network CXRRefineDet is proposed in this paper

Summary

Introduction

Chest X-ray (CXR) is an effective and widely used imaging technique in the diagnosis and screening of lung-related diseases. e imaging principle and structure of chest radiographs are complex, which requires professional radiologists to spend a lot of time to observe carefully. Medical research [1, 2] shows that postprocessing of medical images by using a computer-aided diagnosis (CAD) system can effectively reduce the initial screening of chest radiographs and improve the accuracy of lesion screening. Benefiting from the rapid development of the field of artificial intelligence [3], many researchers have proposed lots of automatic diagnosis methods by combining deep learning technology with imaging examination technology to reduce the workload of radiologists and the possibility of misdiagnosis [4, 5]. Arnaud proposed a new computer-aided detection lung nodule system using multiview convolutional networks (ConvNets) to reduce the false positives of the CAD system [8]. The segmentation model may be inaccurate for the Journal of Healthcare Engineering segmentation of the edge part of the lesion and the small lesion area in practical applications since the spatial dimension of a chest X-ray is usually 2000 × 3000 pixels and the local lesion area is relatively small, which makes the detection more difficult and requires the doctor to spend more time to make further judgments

Methods

Results

Conclusion