Multi-Classification Model for Distinguishing Covid-19 from Different Lung Diseases based on Deep Learning Algorithms | ||
| Labyrinth: Fayoum Journal of Science and Interdisciplinary Studies | ||
| Volume 1, Issue 2, October 2023, Pages 1-10 PDF (1.59 M) | ||
| Document Type: Original full papers (regular papers) | ||
| DOI: 10.21608/ifjsis.2023.211165.1021 | ||
| Authors | ||
| Mohammed Abd-Almoneim Al-Salamony* 1; Ahmed Ismail2; Hazem Elbakry3 | ||
| 1Information Systems Department, Faculty of Computers and Artificial Intelligence, Fayoum University, Fayoum, Egypt | ||
| 2Department of Information Systems, Faculty of Computers and Artificial Intelligence, Fayoum University, El Fayoum 63514, Egypt | ||
| 3Department of Information Systems, Faculty of Computers and Information Mansoura University, El-Mansoura 35516, El-Dakahleya, Egypt | ||
| Abstract | ||
| The Corona-Virus is a worldwide pandemic classified as one of the scariest viruses, according toar the World Health Organization (WHO). That's because of its effect on the person's lungs, which causes high deaths. Among the vital effectiveness indicators for identifying some diseases, including the coronavirus, are computerized tomography (CT) scans and chest X-rays. Data heterogeneity between X-ray and CT biomarkers makes the learning capability of the models more challenging. Furthermore, they utilize multistage for diagnosing COVID-19 from some lung diseases. Hence, the proposed solution behind this research is to leverage form deep learning architecture for applying many classification models to resolve these problems using a fusion of two images that can identify COVID-19, pneumonia, and lung cancer in a single learning procedure. Firstly, patches are extracted from multimodal images by learning every patch using a convolutional neural network (CNN) to address these issues. Then, the available multimodal features are combined for further learning using the AlexNet classifier, CNN classifier, and the Deep Feature Concatenation (DFC) mechanism. Where all the learned features are combined using a straightforward CNN. Finally, the experimental results demonstrated that proposed AlexNet + DFC + CNN exceeded comparable work already done with 98.47 % accuracy | ||
| Keywords | ||
| Lung diseases; Machine learning; Image processing; Patching; CT scan X-ray images | ||
|
Statistics Article View: 239 PDF Download: 187 |
||
