Résumé
The world is going through a global health crisis known as the Covid-19 pandemic. Currently, the outbreak is still evolving in a complicated way with a high spreading speed and new variants appearing constantly. RT-PCR test is preferred to test a patient infected with Covid-19. However, this method depends on many factors such as the time of specimen collection and preservation procedure. The cost to perform the RT-PCR test is quite high and requires a system of specialized machinery for sample analysis. Using deep learning techniques on medial images provides promising results with high accuracy with recent technological advancements. In this study, we propose a deep learning method based on CasCade R-CNN ResNet-101 and CasCade R-CNN EfficientNet in a big data processing environment that accelerates the detection of Covid-19 infections on chest X-rays. Chest X-ray can quickly be performed in most medical facilities and provides important information in detecting suspected Covid-19 cases in an inexpensive way. Experimental results show that the classification of lung lesions infected with Covid-19 has an accuracy of 96% and mAP of 99%. This method effectively supports doctors to have more basis to identify patients infected with Covid-19 for timely treatment. © 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.