Prediction and Detection of COVID-19 from Chest X-Rays using Transfer Learning based Deep Convolutional Neural Networks

Abstract

With the ongoing outbreak of the COVID-19 global pandemic, the research community still struggles to develop early and reliable prediction and detection mechanisms for this infectious disease. The commonly used RT-PCR test is not readily available in areas with limited testing facilities, and it lags in performance and timeliness. This paper proposes a deep transfer learning-based approach to predict and detect COVID-19 from digital chest radiographs. In this study, three pre-trained convolutional neural network-based models (VGG16, ResNet18, and DenseNet121) have been fine tuned to detect COVID-19 infected patients from chest X-rays (CXRs). The most efficient model is further used to identify the affected regions using an unsupervised gradient-based localization technique. The proposed system uses a classification approach (normal vs. COVID-19 vs. pneumonia vs. lung opacity) using three supervised classification algorithms followed by gradient-based localization. The training, validation and testing of the system are performed using 21165 CXR images (10192 normal, 1345 pneumonia, 3616 COVID-19, and 6012 lung opacity). Simulation and evaluation results are presented using standard performance metrics, viz, accuracy, sensitivity, and specificity.

A semantic contour based segmentation of lungs from chest x‐rays for the classification of tuberculosis using Naïve Bayes classifier

“…Subsequently, Naive Bayes classifier can predict the presence of TB on chest x-rays. These features provide better accuracy and specificity values than previously existing methods proposed by Alfadhli et al, 7 Islam et al, 17 and Hwang et al 18 when tested on the Shenzhen dataset. 25 In this research article, a balanced training model is developed using the optimal feature vector of dimension 12.…”

“…On comparison, the proposed algorithm has achieved better AUC and specificity than all the earlier existing algorithms proposed by Islam et al, 17 Stefan Jaeger et al, 12 Santosh et al 19 However, Stefan Jaeger et al 12 has reported better sensitivity than the proposed method due to better quality images and manual optimization techniques. In comparison, the proposed work is verified on…”

“…The local classifier trained with textural features whereas the global classifier is trained with GIST descriptors and finally both the classifiers are fused together to attain higher accuracy. Islam et al 17 and Hwang et al 18 considered Deep Convolutional Neural Networks for the development of TB screening system. Although these methods achieve high accuracy, But hampers with memory allocation problems.…”

Inception‐YOLO: Computational cost and accuracy improvement of the YOLOv5 model based on employing modified CSP, SPPF, and inception modules

“…Techniques for detecting objects have considerably advanced since convolutional neural networks (CNNs) were introduced and are being used in enormous applications and fields like pedestrian detection [17], healthcare [18], and virtual assistants [19]. The ability to recognize objects in pictures and videos is one of the core challenges in computer vision, and it is connected to a wide range of applications, such as face recognition [20], self-driving cars [21], traffic and vehicle detection [22], natural language processing [23], agriculture [24], medical image analysis [25,26] etc.…”

Joint classification and segmentation for an interpretable diagnosis of acute respiratory distress syndrome from chest x-rays

“…Islam et al. employed an ensemble of several models, including Alexnet, Resnet, and VGG 20 . The final result outperformed the use of classifiers for detecting cardiomegaly and tuberculosis.…”