Classification of COVID-19 and Pneumonia from Chest X-ray Images using Deep Learning Techniques

Abstract

Abstract:  A Covid-19 diagnosis utilizing nasopharyngeal swabs and RT-PCR has a low positive rate. Chest X-rays are crucial for early diagnosis of COVID-19, normal and pneumonia. Its symptoms differed from the common cold, influenza, and healthy people. COVID-19 is a major hazard to global health. The diseases are detected through the chest x-ray image dataset by using image classification for deep learning techniques. The image classification for deep learning techniques recognizes the image data and generates the categorized output. As deep neural networks perform the most essential aspect of medical image recognition, pre-processing of the raw image, which are converted into an understandable format by models are required. The models trained for this research include pre-trained CNN models such as VGG, Xception and Dense Net versions. The proposed model’s performance validation is summarised in terms of accuracy, precision, recall, F1 score and AUC values that can aid in early diagnosis and differentiate COVID-19 from other kinds of pneumonia when all the deep learning classifiers and performance parameters were considered, the DenseNet121 achieved the highest model classification of accuracy for COVID-19 at 100%, the Xception of the normal class achieved 95.49% and the DenseNet201 achieved the highest for the viral pneumonia class at 97.14%. Additionally, the suggested method is useful and aids medical professionals in recognizing disorders from chest X-ray images. Even though our architectures are easier to use, the performance of the classifiers is used to prove that the therapies are effective and efficient.

 Index Terms: COVID-19, Pneumonia, CNN, Deep Learning Techniques, chest X-Ray Images, classification.