Please use this identifier to cite or link to this item:
http://148.72.244.84/xmlui/handle/xmlui/15836
Title: | The Use of Convolution Neural Networks to Classify Viral Pneumonia and COVID-19 by Using Chest X-ray Images |
Authors: | Mayssam Alwan Hasson Taha Mohammed Hassan Alaa Jalal Abdullah Sarah Mohemmed Fawzi Hussein |
Keywords: | Convolution Neural Networks, Chest CT scan image, COVID-19 Diagnosis using Deep Learning, Pneumonia, Deep Learning. |
Issue Date: | 1-Jan-2024 |
Publisher: | university of diyala |
Abstract: | The novel coronavirus outbreak reached pandemic status in March-2020. Since then, many countries have collaborated in the fight against COVID-19. The main objective of these governments is the rapid and effective identification of COVID-19-positive patients. While many molecular tests currently exist, not all hospitals have immediate access to these. However, X-rays, which are easily accessible in the majority of hospitals, provide different ways to detect COVID-19. This article discusses the use of neural networks for the classification of radiographic images of patients with pneumonia and COVID-19. Precision, Recall, and F1- score were used to select the best resizing parameters and adaptive equalization of the brightness histogram of images, as well as the optimal architecture of the neural network and its hyperparameters. The high values of these classification quality metrics obtained accuracy (97%) for patients with COVID-19, and accuracy (99%) for patients with pneumonia. These results strongly indicate a reliable differentiation of radiographic images. This opens up the possibility of creating a model with good predictive ability without involving ready-made complex models and preliminary training on third-party data. |
URI: | http://148.72.244.84/xmlui/handle/xmlui/15836 |
ISSN: | 2958-4612 |
Appears in Collections: | مجلة ديالى للعلوم الاكاديمية / Academic Science Journal (Acad. Sci. J.) |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
19-706.pdf | 861.1 kB | Adobe PDF | View/Open |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.