COVID-19 Identification Using Deep Capsule Network: A Perspective of Super-Resolution CNN on Low-Qality CXR Images

Chest X-ray has become a useful method in the detection of coronavirus disease-19 (COVID-19). Due to the extreme global COVID-19 crisis, using the computerized diagnosis method for COVID-19 classification upon CXR images could significantly decrease clinician workload. We explicitly addressed the issue of low CXR image resolution by using Super-Resolution Convolutional Neural Network (SRCNN) to effectively reconstruct high-resolution (HR) CXR images from low-resolution (LR) CXR correspondents. Then, the HRCXR images are fed into the modified capsule network to retrieve distinct features for the classification of COVID-19. We demonstrate the proposed model on a public dataset and achieve ACC of 97.3%, SEN of 97.8%, SPE of 96.9%, and AUC of 98.0%. This new conceptual framework is proposed to play a vital task in the issue facing COVID-19 and related ailments. © 2021 ACM.

Improved convolutional neural multi-resolution wavelet network for COVID-19 pneumonia classification

Fast and early detection of infected patient is the most paramount step necessary to curb the spread of the COVID-19 disease. Radiographs have perhaps presented the fastest means of diagnosing COVID-19 in patients. The well-known standard for COVID-19 test requires a standard procedure and usually has low sensitivity. Previous studies have adopted various AI-based methods in detecting COVID-19 using both chest tomography and chest x-ray. In this study, the goal is to propose an enhanced convolutional neural multi-resolution wavelet network for COVID-19 pneumonia diagnosis. Our proposed model is a convolutional neural network integrated discrete wavelet transform of four level decomposition multiresolution analysis robust to handle few dataset which is very paramount due to the fast emergence of COVID-19. We evaluated our model based on three categories of public dataset of chest x-ray and chest tomography images. Our proposed model achieves 98.5% accuracy, 99.8% sensitivity, 98.2% specificity, and 99.6% AUC for multiple class categories with less training parameters. The results of this study show that our method achieves state-of-the-art result. © 2021 IEEE.

Enhancing low quality in radiograph datasets using wavelet transform convolutional neural network and generative adversarial network for COVID-19 identification

The coronavirus disease of 2019 (COVID-19) pandemic has caused a global public health epidemic since there is no 100% vaccine to cure or prevent the further spread of the virus. With the ever-increasing number of new infections, creating automated methods for COVID-19 identification of Chest X-ray images is critical to aiding clinical diagnosis and reducing the time-consumption for image interpretation. This paper proposes a novel joint framework for accurate COVID-19 identification by integrating an enhanced super-resolution generative adversarial network with a noise reduction filter bank of wavelet transform convolutional neural network on both Chest X-ray and Chest Tomography images for COVID-19 identification. The super-resolution utilized in this study is to enhance the image quality while the wavelet transform Convolutional Neural Network architecture is used to accurately identify COVID-19. Our proposed architecture is very robust to noise and vanishing gradient problem. We used public domain datasets of Chest x-ray images and Chest Tomography to train and check the performance of our COVID-19 identification task. This experiment shows that our system is consistently efficient by accuracy of 0.988, sensitivity of 0.994, and specificity of 0.987, AUC of 0.99, F1-score of 0.982 and 0.989 for precision using the Chest X-ray dataset while for Chest Tomography dataset, an accuracy of 0.978, sensitivity of 0.981, and specificity of 0.979, AUC of 0.985, F1-score of 0.961 and precision of 0.980. These performances have also outweighed other established state-of-the-art learning methods. © 2021 IEEE.

The capability of multi resolution analysis: A case study of COVID-19 diagnosis

The outbreak of COVID-19 in more than 150 nations across the globe is severely impacting the health of people worldwide. A more reliable way to curb the spread of COVID-19 is early detection of infected patients for temporary isolation and care. Image base detection of COVID-19 presents the quickest way to diagnose patients. Few literatures have shown that chest radiograms of infected COVID-19 patients contain irregular characteristics. From related studies, we investigated the application of wavelet transform multi resolution analysis to detect COVID-19 patients using chest radiography. In this study, we proposed a wavelet based convolutional neural network to handle data scarcity in this era of COVID-19 fast emergence. We only considered four levels of wavelet transform decompositions. We utilized an open source dataset from National Institute Health containing several X-rays of pneumonia related diseases whereas the COVID-19 dataset is collected from Radiology Society North America. From the experimental results, our model achieved sensitivity greater than 90% while the specificity is above 90%. We also show the receiver operating characteristic and precision-recall curves of each decomposition level. Our results show that the performance of our proposed model is encouraging and outperformed previous state-of-the-art models. © 2021 IEEE.