Deep Transfer Learning approach for Classification of Chest Infections in Radiographic X-Ray Images

ABSTRACT

From past two years world is suffering with the COVID-19 pandemic which mainly infecting human lungs. Lung infection can be caused by different viruses and bacteria, diagnosis of Lung infections can be done using Radiography X-ray images. Due to similarity in infections there are high chances that other infections can be falsely considered as COVID-19. Manual chest X-ray diagnosis for COVID-19 requires a radiologist and a time taking process, hence it is not a good choice as the covid-19 can spread in no time from person to person.Hence there is need for automatic process of Covid-19 detection and classification of different chest disease. We worked on developing a deep transfer learning model which will accurately classify various chest infections such as COVID-19, Lung-Opacity, Tuberculosis, Viral Pneumonia, We used transfer learning approach which uses existing deep learning models and adding our layers for classification. This work compares various pre-trained models and also convolutional neural network by considering data set of different infections with total 4526 images belonging to 5 classes for training, 647 images belonging to 5 classes for validation, 746 images for testing. © 2022 IEEE.