JMIR Publications

ABSTRACT

Background:

Coronavirus disease has explosively spread globally since the early January of 2020. With the millions of the death rate of individuals, it is essential for an automated system to be utilized for aiding the clinical diagnosis and reduce time consumption for the image analysis.

Objective:

Our aim is to rapidly develop an automated AI model to diagnose COVID-19 in CXR images and differentiate COVID-19 from healthy and other pneumonia.

Methods:

This article presents a GAN-based deep learning application in precisely regaining high-resolution (HR) CXR images from low-resolution (LR) CXR correspondents for COVID-19 identification. Respectively, using the building block of generative adversarial network (GAN), we introduce a modified enhanced super-resolution with generative adversarial network plus (MESRGAN+) to inculcate a connected nonlinear mapping collected from noise-contaminated low-resolution input images to produce deblurred and denoised HR images. As opposed to the latest trend of increasing network elaboration and depth to advance imaging performance, we incorporated an enhanced VGG19 fine-tuned twin network with wavelet pooling strategy in order to extracts distinct features for COVID-19 identification. The qualitative results establish that the proposed model is robust and reliable for COVID-19 screening.

Results:

We demonstrate the proposed enhanced siamese fine-tuned model with wavelet pooling strategy and modified enhanced super-resolution GAN plus based on low quality images for COVID-19 identification on a publicly available dataset of 11,920 samples of chest x-ray images, each having 2,980 cases of COVID-19 CXR, healthy, viral and bacterial cases for our four-class classification. Furthermore, we performed binary classification of COVID-19 verse healthy cases. The proposed method achieves accuracy of 98.8%, precision of 98.6%, sensitivity of 97.5%, specificity of 98.9%, F1-score of 97.8% and ROC AUC of 98.8% for the multi- class task while for the binary class, the model achieved accuracy of 99.7%, precision of 98.9%, sensitivity of 98.7%, specificity of 99.3%, F1-score of 98.2% and ROC AUC of 99.7%.

Conclusions:

Our method obtained state-of-the-art (SOTA) performance, according to experimental results, which is helpful for COVID-19 screening. This new conceptual framework is proposed to play an influential task in the issue facing COVID-19 examination and other ailments, using CXR datasets.