Publication

Abstract : The global surge of COVID-19, an unprecedented viral outbreak originating in China in 2019, method, real-time reverse transcriptase-polymerase chain reaction (RT-PCR), though widely used, is critiqued for its lower sensitivity and prolonged detection time. This necessitates the development of more sensitive, accurate, and rapid screening techniques for mass application. Recent research suggests that chest X-ray images (CXRIs) can offer heightened sensitivity in COVID-19 detection. This paper proposes an image processing and decomposition-based approach to enhance the performance of COVID-19 detection from CXRIs. Leveraging the Two-dimensional Littlewood Paley Empirical Wavelet Transform (2D-LWP-EWT), the pre-processed image is decomposed into sub-band images (SBIs). Features extracted from these SBIs undergo post-processing (concatenation, normalization, selection, and singular value decomposition) before classification by a support vector machine (SVM). With impressive parameters 98.34% accuracy, 100% sensitivity, 96.67% specificity, 97.50% precision, and 98.57 F-Score obtained through 10-fold cross-validation (FCV), this method outperforms existing COVID-19 detection techniques.