A Hybrid Metaheuristic Aware Modified Mobile Net with Enriched Feature Extraction for Covid-19 Severity Detection and Classification

• Published in: Wireless personal communications Vol. 136; no. 2; pp. 1047 - 1077
• Main Authors: Rao, G. V. Eswara, Rajitha, B.
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.05.2024; Springer Nature B.V
• Subjects: Algorithms; Classification; Communications Engineering; Computer Communication Networks; COVID-19; Engineering; Feature extraction; Generative adversarial networks; Heuristic methods; Hybrid systems; Medical imaging; Networks; Optimization; Recurrent neural networks; Signal,Image and Speech Processing; Viral diseases; COVID-19; Generative adversarial networks; Severity; MobileNet; Artificial Ecosystem optimizer and PYTHON
• ISSN: 0929-6212; 1572-834X
• DOI: 10.1007/s11277-024-11315-9

There has been severe illness and mortality globally as a result of the COVID-19 pandemic brought on by the SARS-CoV-2 virus. As the virus continues to spread, it has also undergone mutations leading to the emergence of severity among patients. This severity has raised concerns due to their potential to be more transmissible, cause more severe disease, and escape vaccine-induced immunity. In this study, the goal is to create a technique for identifying and classifying COVID-19 severity using image data. Then the input image is preprocessed using data standardization and image augmentation. The Conditional Generative adversarial networks (CGAN) model is used for feature extraction and the classification is achieved through a novel modified MobileNet. Moreover, a modified Mobile Net is also enhanced through a Hybrid Artificial Eco-System with Monarch Butterfly Optimization (HAEMBO), which is the combination of Artificial Ecosystem optimizer (AEO) and Monarch Butterfly Optimization (MBO). The HAEMBO algorithm enriches the performance of the modified MobileNet. The Recurrent Neural Network (RNN) is employed for severity score detection. The PYTHON platform is used for implementation. The results demonstrated that the proposed method is able to accurately detect and classify COVID-19 severity.