Related Experiment Video
Updated: Oct 10, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.0K
A novel Gray-Scale spatial exploitation learning Net for COVID-19 by crawling Internet resources
Mohamed E ElAraby1, Omar M Elzeki2,3, Mahmoud Y Shams4
1Faculty of Computers and Artificial Intelligence, Beni-Suef University, Beni-Suef 62511, Egypt.
Biomedical Signal Processing and Control
|December 13, 2021
Summary
Scientists developed a Gray-scale Spatial Exploitation Net (GSEN) for COVID-19 detection using Chest X-ray (CXR) images. This deep neural network, trained on web-crawled data, achieved high accuracy, outperforming transfer learning models.
Area of Science:
- Medical Imaging
- Computer Vision
- Artificial Intelligence
Background:
- The COVID-19 pandemic necessitates accurate and rapid diagnostic tools.
- Chest X-ray (CXR) imaging is a primary method for detecting COVID-19, but often lacks detailed information.
- Computer vision techniques can enhance the analysis of grayscale CXR images for improved diagnostic capacity.
Purpose of the Study:
- To develop an efficient deep neural network for COVID-19 detection from CXR images.
- To establish a framework for continuously updating CXR datasets using web crawling.
- To evaluate the performance of the proposed network against established transfer learning models.
Main Methods:
- Designed a Gray-scale Spatial Exploitation Net (GSEN), a lightweight and fast-learning deep neural network.
- Implemented a web crawling framework to construct and continuously update a CXR image dataset.
- Conducted comprehensive evaluations comparing GSEN with pre-trained models like Google-Net, VGG-19, Res-Net 50, and Alex-Net.
Main Results:
- The GSEN achieved high accuracy: 95.60% for two-class (COVID-19 vs. non-COVID-19) and 92.67% for three-class classification.
- Web crawling demonstrated a positive correlation between dataset size and accuracy improvement.
- The proposed GSEN outperformed recent transfer learning approaches in accuracy metrics.
Conclusions:
- The developed GSEN offers an efficient and accurate method for COVID-19 detection using CXR images.
- Web crawling provides a viable strategy for building and maintaining dynamic, large-scale medical image datasets.
- The study highlights the potential of specialized deep learning models and data augmentation techniques in combating infectious diseases.
