Related Experiment Video
Updated: Sep 24, 2025
09:03
Nasal Brushing Sampling and Processing Using Digital High Speed Ciliary Videomicroscopy – Adaptation for the COVID-19 Pandemic
Published on: November 7, 2020
5.0K
CrodenseNet: An efficient parallel cross DenseNet for COVID-19 infection detection
Jingdong Yang1, Lei Zhang1, Xinjun Tang2
1School of Optical-Electrical and Computer Engineering, University of Shanghai for Science and Technology, Shanghai 200093, China.
Biomedical Signal Processing and Control
|May 9, 2022
Summary
This study introduces CrodenseNet, a novel Convolution Neural Network (CNN) for improved COVID-19 detection. CrodenseNet enhances diagnostic accuracy and generalization, aiding clinicians in prompt identification of COVID-19 infection.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Computer Science
Background:
- Convolutional Neural Networks (CNNs) show promise in COVID-19 detection.
- Existing CNN models often suffer from low sensitivity and poor generalization.
Purpose of the Study:
- To propose an effective CNN model, CrodenseNet, for enhanced COVID-19 detection.
- To address the limitations of sensitivity and generalization in current CNN-based methods.
Main Methods:
- Developed CrodenseNet, featuring parallel DenseNet Blocks with dilated and traditional convolutions.
- Incorporated cross-dense connections and one-sided soft thresholding for noise filtering and feature interaction.
Main Results:
- CrodenseNet achieved high performance on the COVID-19x dataset.
- Key metrics include precision (0.967), recall (0.967), F1-score (0.973), AP (0.991), and AUC (0.996).
Conclusions:
- CrodenseNet surpasses state-of-the-art models in COVID-19 detection metrics.
- The model supports clinicians in making prompt and accurate diagnoses of COVID-19 infection.
