KNN-WT BASED COVID-19 DETECTION USING CHEST X-RAY BINARY CLASSIFICATION

ABSTRACT

A novel virus commences in Wuhan China in December 2019. It was named as novel coronavirus (nCovid-19) or severe acute respiratory syndrome corona virus-2 (SARS-CoV-2). Due to its zoonotic nature, it had affected animals as well as human beings. The stated virus is spreading at such a rapid rate that it has razed human lives and the global economy. To aid in such pandemic situation, we have proposed a novel neural network-based model for diagnosing coronavirus from a raw chest X-Ray image. The proposed model uses K-Nearest Neighbor (KNN) for classifying the input image. It will support binary classification i.e., COVID effected X-Ray and normal X-Ray. Several collected input images are initially pre-processed using dual-tree complex wavelet transform (DTCWT). Then, feature extraction is executed using mobilenet architecture. Further, image classification is performed using the KNN based model. Lastly, the output is predicted whether it belongs to the Covid-19 class or normal class. For visualizing the effectiveness of the proposed KNN based classifier, parameters such as accuracy, recall, precision, and F1 score are calculated. A comparison is made by calculating the average of all the parameters with existing techniques. Experimental results showed that the proposed KNN-WT model achieves an accuracy of 99%. It outperformed all the existing algorithms. © 2023 Little Lion Scientific.