Automatic diagnosis of CoV-19 in CXR images using haar-like feature and XgBoost classifier

Many researchers and medical practitioners have recently focused on the automatic detection of COVID-19 using chest X-Ray (CXR) images. In the past, several computer-aided diagnostic (CAD) schemes were developed to assist healthcare professionals. These CAD systems using CXR images can speed up the identification process. As a result, we have proposed an automatic diagnosis (CAD-CXR) system by using the haar-like feature and XgBoost classifier. To develop a CAD-CXR system, we have used a pre-processing step and Haar-Like feature descriptor to extract features from chest X-ray images. Afterward, many machine learning classifiers are employed to categorize the images into coronavirus-19 and normal. Feature extraction and classification techniques are evaluated based on detection performance. Several experiments are performed to compare the performance of the proposed method with other classifiers. Our work is evaluated by public datasets with three different train-test splits in terms of average accuracy, F1-score, recall, and precision metrics. Compared to state-of-the-art methods, the XgBoost classifier with the Haar-Like feature obtained excellent performance based on the accuracy of 97.5%, F1-score of 96%, recall of 96%, and precision of 97%. These experimental results conclude that the CAD-CXR system outperforms to detect most of the covid-19 cases. The proposed CAD-CXR approach can be utilized to effectively screen the COVID-19-infected patients using CXR images.