Identification of COVID-19 patients using an intelligent technique

Abstract

Coronavirus disease is considered a serious disease because of its devastating impact on the health and life of the world and its significant impact on the deterioration of the economic and commercial situation in the world. In this paper, we presented a technique for diagnosing the Coronavirus disease based on deep learning and discrete wavelet transform. The discrete wavelet transform used in the pre-processing stage to reduce the complexity witch led to increasing the speed of the proposed technique while the deep learning used in the feature extraction and classification stage. Unlike the other methods in this domain, the proposed technique evaluates the noise impact on the accuracy of the proposed technique by considering two type of the noise namely: pepper & Salt noise and Gaussian Noise. Furthermore, the proposed technique evaluates the effect of the rotation on the accuracy by taken different rotation angles. The analysis of the extensive experiments that carried out on the dataset refer that the proposed technique achieve high accuracy (99%) with less computation time, as well as it robust against the noise and rotation variations.