Classification of covid-19 and other lung diseases from chest x-ray images

There are several lung diseases such as pneumonia, asthma, tuberculosis, and fibrosis. The most recently, coronavirus disease 2019 (COVID-19), is rapidly spreading and cause a pandemic with a many of victims. The standard test method for diagnosis of the disease, described by World Health Organization (WHO), is Real-time reverse transcription polymerase chain reaction (RT-PCR) which takes long from several hours to two days. In addition, considering some shortcoming of the testing by kit, such as limitation in number of kits, and probability to spread the virus during the test procedure depicts a necessary of presenting automatic diagnosis of COVID-19 from medical imaging such as chest X-ray to control this dangerous pandemic. This study aims to develop and test several different deep learning models using Convolutional Neural Network (CNN)-based models as well as vision transformer (ViT) in image classification to automatically diagnose COVID-19 and other kinds of lung diseases using chest X-Ray as input image. In this thesis, different CNN-based deep learning models are proposed. This CNN-based models can be trained to classify chest X-ray images into three classes of COVID-19, normal chest X-ray and other lung diseases. The different proposed models will be trained with three-class balanced dataset which consists of 3000 images, 1000 images for each class. Besides, the binary classification between two classes of COVID-19 and normal chest X-ray is proposed. In addition of using CNN, two different models are trained with three-class dataset. Nine different models and their results is proposed with a comparison of their results. A publicly available dataset to train and test the CNN model is used from Kaggle- COVID-19_Radiography_Dataset. From the experiments, the accuracy of VGG16 model is 93.44 % and ViT is 92.33 %