Custom YOLOV4 object detection model for COVID-19 diagnosis and cell counting

After different consecutive waves, the pandemic phase of COVID-19, which started in Wuhan, China, seems to have no imminent end worldwide transmission of the SARS coronavirus has prompted the World Health Organization to proclaim a worldwide pandemic (known as COVID-19). Its effects on regular life...

Full description

Saved in:
Bibliographic Details
Main Authors: Ali, Noor Najah, Hameed, Aseel, Al-Naji, Ali
Format: Conference Proceeding
Language:English
Subjects:
Artificial intelligence
Blood cells
Coronaviruses
COVID-19
Diagnostic systems
Disease transmission
Electron microscopy
Health surveillance
Image processing
Machine learning
Medical imaging
Object recognition
Pandemics
Surveillance systems
Viral diseases
Viruses
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:After different consecutive waves, the pandemic phase of COVID-19, which started in Wuhan, China, seems to have no imminent end worldwide transmission of the SARS coronavirus has prompted the World Health Organization to proclaim a worldwide pandemic (known as COVID-19). Its effects on regular life and health have been devastating. In order to stop the spread of the disease and provide immediate care for infected patients, Rapid identification of positive cases is essential. The demand for diagnostic aids has grown. Recent advances in medical imaging show that these pictures reveal important information about the COVID-19 virus. The current shortage of specialists can be alleviated in part through the use of cutting-edge user artificial intelligence technologies and medical imagery in making correct diagnoses of this condition. The small size of the virus cells makes it difficult to distinguish them visually with the naked eye. Therefore, the transmission electron microscope with different magnifications was used to detect and image the virus cells. In recent years, medical diagnoses have significantly benefited from the widespread use of machine learning-based image processing technologies and have proven to be an effective method for helping diagnostic professionals to identify subtle lesions in images accurately. This work presents a customized computer-assisted automatic detection system specifically focused on quantifying and counting viral cells detected in blood samples from patients using electron microscopy images to detect infected blood cells. The average accuracy of the proposed custom model is 86.5% mAP, which makes it suitable for the upcoming COVID-19 surveillance system.
ISSN:0094-243X
1551-7616
DOI:10.1063/5.0191735