CovidGAN: Data Augmentation Using Auxiliary Classifier GAN for Improved Covid-19 Detection.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Infectious Diseases

Background:

• Coronavirus disease (COVID-19), caused by SARS-CoV-2, poses global health and economic challenges.
• Chest X-rays (CXRs) are vital for diagnosing COVID-19, revealing characteristic abnormalities.
• Deep learning models, particularly Convolutional Neural Networks (CNNs), show promise for automated COVID-19 detection from CXRs.

Purpose of the Study:

• To address the scarcity of training data for deep learning models in early COVID-19 detection.
• To develop a method for generating synthetic COVID-19 positive chest X-ray images.
• To evaluate the effectiveness of synthetic images in improving CNN-based COVID-19 detection accuracy.

Main Methods:

• Development of an Auxiliary Classifier Generative Adversarial Network (ACGAN) named CovidGAN to generate synthetic CXR images.
• Training and testing CNN models with and without the inclusion of synthetic CXR data.
• Comparative analysis of classification accuracy between models.

Main Results:

• CNN classification alone achieved 85% accuracy in detecting COVID-19 from chest X-rays.
• Incorporating synthetic images generated by CovidGAN boosted CNN model accuracy to 95%.
• The study demonstrates the utility of synthetic data for enhancing diagnostic performance.

Conclusions:

• CovidGAN provides a viable method for generating realistic synthetic chest X-ray images for COVID-19 detection.
• Synthetic data generated by CovidGAN significantly improves the accuracy of deep learning models for COVID-19 diagnosis.
• This approach can accelerate the development of robust AI-powered radiological systems for infectious disease detection.