Covid-19 Detection Based on Chest X-Ray Images Using DCT Compression and NN

Covid-19 is a highly contagious virus spreading all over the world. It is caused by SARS-CoV-2. virus. Some of the most common symptoms are fever, cough, sore throat, tiredness, and loss of smell or taste. There are two types of tests for COVID-19: the PCR test and the antigen test. Automatic detection of Covid-19 from publicly available resources is essential. This paper employs the commonly available chest x-ray (CXR) images in the classification of Covid-19, normal and viral pneumonia cases. The proposed method divides the CXR images into subblocks and computes the Discrete Cosine Transform (DCT) for every subblock. The DCT energy compaction capability is employed to produce a compressed version for each CXR image. Few spectral DCT components are incorporated as features for each image. The compressed images are scanned by average pooling windows to reduce the dimension of the final feature vectors. A multilayer artificial neural network is employed in the 3-set classification. The proposed method achieved an average accuracy of 95 %. While the proposed method achieves comparable accuracy relative to recent state-of-the-art techniques, its computational burden and implementation time is much less.