Detection and identification of COVID -19 based on chest medical image by using convolutional neural networks
Jothi V. Pranav, R. Anand, T. Shanthi, K. Manju, S. Veni, S. Nagarjun
Abstract
Covid-19 pandemic has caused major out-break all around the world. This pandemic out-break requires lot of testing, which is a tedious process. Deep learning is a successful method that has evolved in image category in the past few years. In this work to detects the presence of coronavirus by using deep learning approach. Here, convolutional neural networks with specific focus on to classify Covid-19 chest radiography images. The database comprises Covid-19, normal and viral pneumonia chest X-ray images with 800 different samples under each class. We evaluated the model on 500 images and the networks has achieved a sensitivity rate of 95% and specificity rate of 97%. The DenseNet121 Architecture performed slightly better, compared to other state of art networks. The performance achieved by the method proposed is very encouraging and the accuracy rates can be improved further with larger datasets. Apart from sensitivity and specificity rates, the proposed model is also compared on receiver operating characteristic (ROC), and area under the curve (AUC) of each model. The model is implemented on the TensorFlow framework with the datasets that are publicly available for research community.