Title: Generating multiclass COVID-19 CT scan images using multi-convolutional conditional GAN based on deep learning techniques

Authors: M. Anusha; P. Kiruthika

Addresses: PG & Research Department of Computer Science, National College (Autonomous), Affiliated to Bharathidasan University, Tamil Nadu, India ' PG & Research Department of Computer Science, National College (Autonomous), Affiliated to Bharathidasan University, Tamil Nadu, India

Abstract: Medical image analysis, particularly for CT scans, plays a crucial role in the diagnosis and management of various diseases, including COVID-19. However, the limited availability of diverse and representative data poses challenges in developing accurate machine-learning models for CT scan analysis. This research proposes a multi convolutional conditional generative adversarial network (MCC-GAN) for generating CT scans of different classes, including normal, COVID-19, pneumonia, Omicron, and Delta. The discriminator and generator architectures are designed, and image normalisation and CLAHE pre-processing techniques are applied. The training process is monitored using loss graphs, and the generated CT scans are visually realistic and diverse. The proposed multi-multi-convolutional conditional GAN-based approach has the potential to overcome data scarcity challenges and improve the robustness of deep learning models for CT scan analysis. Further validation of clinical datasets is warranted to establish the effectiveness of the proposed approach in real-world medical image analysis scenarios.

Keywords: COVID-19 CT scans; multi convolutional conditional GAN; pneumonia; Omicron; Delta; multi-class augmentation; pre-processing techniques.

DOI: 10.1504/IJIEI.2024.137714

International Journal of Intelligent Engineering Informatics, 2024 Vol.12 No.1, pp.1 - 26

Received: 06 Jun 2023
Accepted: 21 Oct 2023
Published online: 02 Apr 2024 *

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