Related Experiment Video
Updated: Jul 8, 2026
11:00
High-throughput Detection of Respiratory Pathogens in Animal Specimens by Nanoscale PCR
Published on: November 28, 2016
12.0K
A Novel Method for COVID-19 Detection Based on DCNNs and Hierarchical Structure.
Yuqin Li1, Ke Zhang1,2, Weili Shi1,2
1School of Computer Science and Technology, Changchun University of Science and Technology, Changchun 130022, China.
Computational and Mathematical Methods in Medicine
|September 12, 2022
Summary
This study introduces a novel deep learning framework using transfer learning and attention networks for accurate COVID-19 detection from chest X-rays. The method enhances feature learning and reduces misclassification, improving diagnostic accuracy for early screening.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Computer-Aided Diagnosis
Background:
- The COVID-19 pandemic necessitates rapid and accurate diagnostic tools.
- Chest X-rays are crucial for COVID-19 diagnosis, but detection models face challenges like subtle lesion differences and limited data.
- Existing deep learning models struggle with insufficient accuracy for effective early screening.
Purpose of the Study:
- To develop an enhanced deep learning framework for improved COVID-19 detection accuracy using chest X-ray images.
- To leverage transfer learning and attention mechanisms to enhance feature representation in deep convolutional neural networks (DCNNs).
- To address the limitations of existing models in detecting subtle lesions and handling data scarcity.
Main Methods:
- A transfer learning strategy is applied to a hierarchical structure of DCNNs.
- Asymmetric pretrained DCNNs are integrated with attention networks to learn discriminative and complementary features.
- A novel cross-entropy loss function with a penalty term is proposed to minimize misclassification errors.
Main Results:
- The proposed framework demonstrated high performance and effectiveness on a COVID-19 dataset.
- Experimental results showed significant improvements compared to state-of-the-art methods.
- The integration of attention networks and the modified loss function enhanced feature learning and reduced misclassifications.
Conclusions:
- The developed framework offers a promising approach for accurate and efficient COVID-19 screening using chest X-rays.
- The study highlights the potential of transfer learning and attention mechanisms in medical image analysis for infectious diseases.
- This method can aid in timely diagnosis and management of COVID-19 cases, contributing to public health efforts.
