Open Access

ARTICLE

Leveraging Convolutional Neural Network for COVID-19 Disease Detection Using CT Scan Images

Mehedi Masud^*, Mohammad Dahman Alshehri, Roobaea Alroobaea, Mohammad Shorfuzzaman

Department of Computer Science, College of Computers and Information Technology, Taif University, P.O. Box 11099, Taif, Saudi Arabia

* Corresponding Author: Mehedi Masud. Email:

Intelligent Automation & Soft Computing 2021, 29(1), 1-13. https://doi.org/10.32604/iasc.2021.016800

Received 12 January 2021; Accepted 11 March 2021; Issue published 12 May 2021

Abstract

In 2020, the world faced an unprecedented pandemic outbreak of coronavirus disease (COVID-19), which causes severe threats to patients suffering from diabetes, kidney problems, and heart problems. A rapid testing mechanism is a primary obstacle to controlling the spread of COVID-19. Current tests focus on the reverse transcription-polymerase chain reaction (RT-PCR). The PCR test takes around 4–6 h to identify COVID-19 patients. Various research has recommended AI-based models leveraging machine learning, deep learning, and neural networks to classify COVID-19 and non-COVID patients from chest X-ray and computerized tomography (CT) scan images. However, no model can be claimed as a standard since models use different datasets. Convolutional neural network (CNN)-based deep learning models are widely used for image analysis to diagnose and classify various diseases. In this research, we develop a CNN-based diagnostic model to detect COVID-19 patients by analyzing the features in CT scan images. This research considered a publicly available CT scan dataset and fed it into the proposed CNN model to classify COVID-19 infected patients. The model achieved 99.76%, 96.10%, and 96% accuracy in training, validation, and test phases, respectively. It achieved scores of 0.986 in area under curve (AUC) and 0.99 in the precision-recall curve (PRC). We compared the model’s performance to that of three state-of-the-art pretrained models (MobileNetV2, InceptionV3, and Xception). The results show that the model can be used as a diagnostic tool for digital healthcare, particularly in COVID-19 chest CT image classification.

---

Keywords

---

Citations