Covid-19 Detection Based on the Fine-Tuned MobileNetv2 Through Lung X-rays
4th International Symposium on Advanced Electrical and Communication Technologies, ISAECT 2021
; 2021.
Article
in English
| Scopus | ID: covidwho-1714068
ABSTRACT
Covid-19 is has become an epidemic, which is affecting millions of people around the world. The common symptoms of Covid-19 are cough and fever, which are very similar to the normal Flu. Covid-19 spreads fast and is devastating for people of all ages especially elderly and people having weak immune system. The standard technique used for Covid-19 detection is real-time polymerase chain reaction (RT-PCR) test. However, RT-PCR is unreliable for Covid-19 detection as it takes long time to detect the disease and it produces considerable number of false positive cases. Therefore, we need to propose an automated and reliable method for Covid-19 detection. Radiographic images are widely used for the detection of various pulmonary diseases such as lung cancer, asthma, pneumonia, etc. We also used chest x-rays for the diagnosis of Covid-19. In this paper, we employed two deep learning models such as SqueezeNet and MobileNetv2 and fine-tuned to check the classification performance. Moreover, we performed data augmentation technique to increase the amount of data and avoid the overfitting of model. We evaluated the performance of the proposed system on standard dataset Covid-19 Radiography dataset that is publicly available. More specifically, we achieved remarkable accuracy of 97%, precision of 95.19%, recall of 100%, specificity of 95%, area under the curve of 98.93%, and F1-score of 97.06% on MobileNetv2. Experimental results and comparative analysis with other existing methods demonstrate that our method is reliable than PT-PCR and other existing state-of-the-art methods for Covid-19 detection. © 2021 IEEE.
Full text:
Available
Collection:
Databases of international organizations
Database:
Scopus
Language:
English
Journal:
4th International Symposium on Advanced Electrical and Communication Technologies, ISAECT 2021
Year:
2021
Document Type:
Article
Similar
MEDLINE
...
LILACS
LIS