Title: Efficient deep transfer learning based COVID-19 detection and classification using CT images

Authors: G. Prabakaran; K. Jayanthi

Addresses: Department of Computer and Information Science, Faculty of Science, Annamalai University, Tamil Nadu, India ' Department of Computer Application, Government Arts College, C-Mutlur, Chidambaram, Tamil Nadu, India

Abstract: This paper develops an intelligent deep transfer learning-driven COVID-19 detection and classification model using CT images. The major aim of the IDTLD-CDCM model is to identify appropriate class labels for the CT images. The IDTLD-CDCM model undergoes initial pre-processing in two levels namely spline adaptive filtering (SAF) based noise removal and contrast enhancement. In addition, the IDTLD-CDCM model involves SqueezeNet as a feature extractor for deriving a useful set of feature vectors. Furthermore, the hop field neural network (HFNN) model is utilised for the classifier of COVID-19 and Non-COVID-19 images. Furthermore, the parameter tuning of the HFNN model is carried out by the use of root mean square propagation (RMSProp). To investigate the improved outcomes of the IDTLD-CDCM approach, a series of simulations are executed and the outcomes are inspected in several aspects. The simulation outcome demonstrated the improved outcomes of the IDTLD-CDCM approach over the recent approaches.

Keywords: COVID-19; deep transfer learning; computed tomography images; medical imaging; decision making; machine learning.

DOI: 10.1504/IJSSE.2024.137073

International Journal of System of Systems Engineering, 2024 Vol.14 No.2, pp.174 - 189

Received: 05 Oct 2022
Accepted: 22 Dec 2022
Published online: 01 Mar 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article