Hybrid deep recurrent neural networks for COVID-19 detection and diagnosis
| Autor: | R S Soundariya, R M Tharsanee, J Nirmaladevi, B Vishnupriya, M Nivaashini |
|---|---|
| Rok vydání: | 2022 |
| Předmět: | |
| Zdroj: | International journal of health sciences. :8551-8564 |
| ISSN: | 2550-696X 2550-6978 |
| Popis: | Corona virus Disease (COVID-19) is an acute pandemic which has put the lives of millions of people worldwide at risk during recent times. There is a high demand to develop effective tools and methods to diagnose the COVID infection in people at an early stage to prevent the spread of the disease to a larger community. This paper aims to provide a systematic method for COVID diagnosis using machine learning and deep learning algorithms. The proposed method Hybrid Deep Recurrent Neural Network (HDRNN) is a fusion of Convolution Neural Networks (CNN) and Long Short-Term Memory-Recurrent Neural Networks (LSTM-RNN) to detect COVID infection efficiently from X-ray samples. CNN is employed in the proposed method primarily to extract the essential features from the X-ray images and LSTM is suitable to classify the COVID affected patients with more fidelity. The dataset used in this work consists of an aggregate of 3470 images including COVID affected and Pneumonia affected samples. The experimental results carried out on the collected dataset with the proposed HDRNN method demonstrated an accuracy of 99.4%, F1 Score 98.7%, Sensitivity of 99.3% and Specificity of 99.2 %. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|