Automatic diagnosis of COVID-19 from CT images using CycleGAN and transfer learning

The outbreak of the corona virus disease (COVID-19) has changed the lives of most people on Earth. Given the high prevalence of this disease, its correct diagnosis in order to quarantine patients is of the utmost importance in the steps of fighting this pandemic. Among the various modalities used for diagnosis, medical imaging, especially computed tomography (CT) imaging, has been the focus of many previous studies due to its accuracy and availability. In addition, automation of diagnostic methods can be of great help to physicians. In this paper, a method based on pre-trained deep neural networks is presented, which, by taking advantage of a cyclic generative adversarial net (CycleGAN) model for data augmentation, has reached state-of-the-art performance for the task at hand, i.e., 99.60% accuracy. Also, in order to evaluate the method, a dataset containing 3163 images from 189 patients has been collected and labeled by physicians. Unlike prior datasets, normal data have been collected from people suspected of having COVID-19 disease and not from data from other diseases, and this database is made available publicly. Moreover, the method's reliability is further evaluated by calibration metrics, and its decision is interpreted by Grad-CAM also to find suspicious regions as another output of the method and make its decisions trustworthy and explainable.& COPY; 2023 Elsevier B.V. All rights reserved.

Automated summary

The outbreak of COVID-19 has had a significant impact on people worldwide. Accurately diagnosing and isolating patients is crucial in fighting this pandemic, and medical imaging, particularly CT imaging, has been a focus of research due to its accuracy and availability. This paper presents a method using pre-trained deep neural networks and a CycleGAN model for data augmentation, achieving state-of-the-art performance with 99.60% accuracy. A dataset of 3163 images from 189 patients, collected from suspected COVID-19 cases, has been publicly made available for evaluation. The method's reliability is further assessed using calibration metrics and the Grad-CAM technique for explaining its decisions.