R-CovNet: hybridized model for automated classification of COVID-19 cases from chest X-ray images

COVID-19 has significantly disrupted our normal way of life for an extended period. Moreover, the COVID-19 situation has galvanized the artificial intelligence (which encompasses machine learning and deep learning) community to create new solutions which can help in the fight against this deadly virus. There is a demand for an early diagnosis system that can provide fast decision and reduction in diagnosis errors. In this paper, a novel hybrid model for classifying chest X-ray digital images for COVID-19 cases is proposed named as R-CovNet. The approach involves extracting image features using a deep convolution neural network (DCNN) architecture and subsequently reducing the dimensionality of these features through the application of Principal Component Analysis (PCA). It is worthy to mention that before extracting the image features, all the images are passed through the noise and contrast enhancement mechanisms. Finally, random forest (RF) classifier is developed to classify the COVID-19 patients from the chest X-ray digital images. The model has been validated on publicly available image data and the best performance is obtained with an accuracy of 99.33%. The superiority of the model over compared methods is well evident from the results.