Classification of COVID-19 in chest X-ray images using DeTraC deep convolutional neural network

Asmaa Abbas; Mohammed Abdelsamea; Mohamed Gaber

Class decomposition [14] has been proposed with the aim of enhancing low variance 37 classifiers facilitating more flexibility to their decision boundaries. In this paper, we 38 adapt and validate DeTraC [15] for the classification of COVID-19 in chest x-ray images 39 1 . This is by adding a class decomposition layer to the pre-trained models. The class 40 decomposition layer aims to partition each class within the image dataset into several 41 sub-classes and then assign new labels to the new set, where each subset is treated as an 42 independent class, then those subsets are assembled back to produce the final 43 predictions. For the classification performance evaluation, we used images of chest x-ray 44 collected from several hospitals and institutions. The dataset provides complicated 45 computer vision challenging problems due to the intensity inhomogeneity in the images 46 and irregularities in the data distribution. Then we apply the class-decomposition layer of DeTraC to simplify the local structure 52 of the data distribution. In the second phase, the training is accomplished using a 53 sophisticated gradient descent optimisation method. Finally, we use the 54 class-composition layer of DeTraC to refine the final classification of the images. As pre-trained CNN model using the collected chest X-ray image dataset. We used the 64 off-the-shelf CNN features of pre-trained models on ImageNet (where the training is 65 accomplished only on the final classification layer) to construct the image feature space. 66 However, due to the high dimensionality associated with the images, we applied PCA to 67 project the high-dimension feature space into a lower-dimension, where highly 68 correlated features were ignored. This step is important for the class decomposition to 69 produce more homogeneous classes, reduce the memory requirements, and improve the 70 efficiency of the framework. . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.