Automated X-Ray Image Classification for COVID-19 Detection Using DenseNet-121 and SVM

This paper provides an algorithm to detect COVID-19 from the chest X-Ray images using Two-class classification with DenseNet-121 and Support Vector Machine. For effective identification of patterns suggestive of COVID-19 in the provided X-ray images, we leverage the DenseNet-121 model due to its feature extracting prowess, complemented by the SVM classifier to accurately classify the images between the COVID-19 to non-COVID-19 positive cases. The method delivers a high accuracy, which is more than 93% for accuracy, precision, recall, and F1-score, which makes it efficient in the COVID-19 quick screening. The Improved DenseNet-121 + SVM was compared with other deep learning models like ResNet50, VGG16, and confirmed that the proposed model has better efficiency in classifying COVID-19 cases. The performance of the model is checked with a five fold cross validation to ensure consistency and reliability on different sets of data not only in terms of performance with the IMDB data but also with the test data. The working paper is relevant to the current research practices in an effort to construct machine diagnostic systems that can be used by practicing healthcare professionals for identifying COVID-19 in the early stages, thus enhancing outcomes in patients' treatment.