A New Machine Learning-Driven Approach for the Diagnosis of COVID-19: An Ultra Covix Model

Abstract The rigorous clinical prognosis is ambiguous due to the ongoing global crisis caused by different mutant variations of the prevailing COVID-19 pandemic. Hitherto, various clinical prognosis imaging techniques are suggested to medical practitioners to identify COVID-19 contracted individuals. Herein, we demonstrate an efficient tool aiding ultrasound imaging technique backed by machine learning strategies, which help diagnose COVID-19 infected cases more accurately and efficiently. The latter approach complements CT and chest X-ray imaging methods. Accordingly, our novel method employs gradient mapping and distinct haralick features using the image database (705 Ultrasound Images). We also propose a vivid technique that assists in diagnosing COVID-19 contaminated individuals by examining ultrasound pictures to identify novel coronavirus. The test set of the precision score is analyzed in the light of attainment results viz., accuracy, confusion matrix, and ROC curve by utilizing the GitHub repository, which conforms to their endorsed ultrasound images. Various algorithms are used to examine test sets accompanying 211 clinical image data for classification performance. Interestingly, the article reveals that the multiple classification accuracy of the proposed model has achieved 98.1% accuracy between the COVID-19, normal, and Pneumonia ultrasound image database.