Cognitive Analytics and Reinforcement Learning

7COVID-19 Detection through CT Scan Image Analysis: A Transfer Learning Approach with Ensemble Technique

P. Padmakumari*, S. Vidivelli and P. Shanthi

School of Computing, SASTRA University, Thanjavur, Tamil Nadu, India

Abstract

The COVID-19 pandemic has had a profound impact on the world due to the highly contagious nature of the virus. To effectively control its spread, identifying and treating patients in the early stages of the disease is crucial. Although reverse transcription-polymerase chain reaction (RT-PCR) is the standard method for COVID-19 detection, it can be prone to errors and time-consuming. To address this issue, deep learning techniques and computed tomography (CT) scans are being used for automated COVID-19 detection. This method involves using the Transfer Learning technique VGG-16 to extract essential features from 2482 CT scan images. Prior to extraction, the images are enhanced through Contrast-limited adaptive histogram equalization (CLAHE). The extracted features are then utilized to train various Machine Learning (ML) models, including Support Vector Machine, Logistic Regression, and Gradient Boosting models. To further improve performance, the identification results from each model are combined in an ensemble learning model using the Hard voting ensemble. This approach achieved an accuracy rate of 98.927%.

Keywords: COVID-19, CLAHE, ResNet, DenseNet, VGG-16, inception, feature extraction, SVM

7.1 Introduction

At the end of December 2019, Wuhan, ...