COVID-19 Detection in CT/X-ray Imagery Using Vision Transformers

Abstract

The steady spread of the 2019 Coronavirus disease has brought about human and economic losses, imposing a new lifestyle across the world. On this point, medical imaging tests such as computed tomography (CT) and X-ray have demonstrated a sound screening potential. Deep learning methodologies have evidenced superior image analysis capabilities with respect to prior handcrafted counterparts. In this paper, we propose a novel deep learning framework for Coronavirus detection using CT and X-ray images. In particular, a Vision Transformer architecture is adopted as a backbone in the proposed network, in which a Siamese encoder is utilized. The latter is composed of two branches: one for processing the original image and another for processing an augmented view of the original image. The input images are divided into patches and fed through the encoder. The proposed framework is evaluated on public CT and X-ray datasets. The proposed system confirms its superiority over state-of-the-art methods on CT and X-ray data in terms of accuracy, precision, recall, specificity, and F1 score. Furthermore, the proposed system also exhibits good robustness when a small portion of training data is allocated.

SCovNet: A skip connection-based feature union deep learning technique with statistical approach analysis for the detection of COVID-19

“…In the event of an unbalanced small dataset, a modified version of the ResNet structure, that is, SCovNet, was used for testing. Performance metrics such as accuracy, specificity sensitivity, precision, F1-score and Matthews’s correlation coefficient are calculated using classifier-generated confusion matrix [42] , [43] , [44] , [45] , [46] . The accurately detected cases in the confusion matrix’s diagonal region are used to calculate the effectiveness of the deep learning classifier [44] , [45] .…”

Contrasting EfficientNet, ViT, and gMLP for COVID-19 Detection in Ultrasound Imagery

“… where , , , and are the true positive, true negative, false positive, and false negative values, respectively. These evaluation metrics are used by several similar works in the literature [ 8 , 13 , 34 , 41 ]; thus, using them gives the ability to compare our results with those of the other studies and to be consistent with the assessment procedures of medical diagnostic systems [ 58 ]. These metrics are calculated from a confusion matrix.…”

“…The literature on COVID-19 reports several methods for the analysis of medical images such as of CT images [ 35 , 36 , 37 , 38 , 39 , 40 , 41 ], X-ray [ 42 , 43 , 44 , 45 , 46 , 47 ], and LUS [ 7 , 8 , 9 , 10 ]. For instance, Silva et al [ 35 ] proposed a voting-based approach, where the images from a given patient are classified using a group in a voting system.…”

Recent Progress in Transformer-based Medical Image Analysis

“…The proposed model is composed of three modules, which are a pretext channel module, a transformer-based transfer module, and a downstream channel module. Rahhal and colleagues [101] proposed a customized framework to classify X-ray and CT images [102,103]. The developed framework is symmetric and consists of two transformers.…”

“…The localization model UTRAD proposed by Chen et al [195], as shown in Table 3, is designed for both localization and detection using the same datasets [196,197,198]. The classification framework proposed by Rahhal and colleagues [101], as illustrated in Table 1, is also designed for image detection and the datasets used for model evaluation remain unchanged [102,103]. Captioning.…”