Accurate Prediction of COVID-19 using Chest X-Ray Images through Deep Feature Learning model with SMOTE and Machine Learning Classifiers

Rahul Kumar; Ridhi Arora; Vipul Bansal; Vinodh J Sahayasheela; Himanshu Buckchash; Javed Imran; Narayanan Narayanan; Ganesh N Pandian; Balasubramanian Raman

In this work, we have presented the use of ResNet152 and machine learning classifiers for the effective classification of COVID-19. The proposed methodology is trained on two publicly available datasets and has outperformed across all the classes. We also encompassed the SMOTE algorithm for balancing the intra-class variation among the datasets. With the SMOTE based features, machine learning algorithms are applied for final classification leading to the best result obtained by Random Forest with the Accuracy, Sensitivity, Specificity, F1-score and AUC of 0.973, 0.974, 0.986, 0.973, and 0.997 (for Random Forest) and 0.977, 0.977, 0.988, 0.977, and 0.998 (for XGBoost), respectively. Therefore, this approach of using X-ray images and computer-aided diagnosis can be used as a massive, faster and cost-effective way of screening. Also, it brings down the time for testing drastically. To make a clinically effective prediction of COVID-19, training with more massive datasets and testing in the field with a larger cohort can be immensely useful.