COVID-19 Identification Based on Keras DenseNet201 Architecture Model Using CT Image

Abstract

Coronavirus illness 2019 (COVID-19) is a spreading sickness produce by the new Coronavirus named serious basic respiratory disorder Covid 2 (SARS-CoV-2). It has been widely increased worldwide after all the beginning of 2020. many Deep Learning (DL) has been demonstrated for classification, segmentation, and detection takes in medical imaging. The principal investigation in this paper, we applying keras DenseNet-201 transfer learning prototype to classify and eventually obtain an empirical and realistic computer-aided symptomatic prototype. In this inspection, we removed the noises from the computed tomography (CT) images into build a new data source, then used the customize keras DenseNet-201 mould construct on transfer learning to bring out features frequently, and exercised softmax activation classifier to codify the CT images. The model is evaluated dataset from COVID-19-CT Database. Examinations are made across 746 CT figures.in the dataset, contain 349 CT images of patients with COVID-19 and 379 CT images for non-COVID-19. The classification model shows around 90.94% testing accuracy. Comparatively, this approach obtains high precision rate rather than other propose methods which are predicted to use small size of dataset samples.