COVID-19 identification in chest X-ray images on flat and hierarchical classification scenarios
share
The COVID-19 is estimated to have a high impact on the healthcare system. In this context, early diagnosis is crucial for correct treatment in order to possibly reduce the stress in the healthcare system. This study aims to identify pneumonia caused by COVID-19 from other types and also healthy lungs using only CXR images. Our main objective is to classify CXR images in order to identify COVID-19 in the images. Therefore, we have proposed a classification schema considering the following perspectives: i) a multi-class classification scenario (flat classification); ii) hierarchical classification scenario, since we can structure the different kinds of pneumonia in a hierarchy. Given the natural data imbalance in this application domain, we also proposed the use of resampling algorithms in the schema in order to re-balance the classes distribution. We observed that, texture is one of the main visual attributes of CXR images, our classification schema extract features using some well-known texture descriptors and also using a pre-trained CNN model. We also explored early and late fusion techniques in the schema in order to leverage the strength of multiple texture descriptors and base classifiers at once. In order to evaluate the proposed approach, we composed a database, named RYDLS-20, containing multiple CXR images of pneumonia caused by different pathogens as well as CXR images of healthy lungs. The distribution of classes in RYDLS-20 followed a real-world scenario in which some pathogens are more common than others. The proposed classification approach achieved a micro-avg F1-Score of 0.65 using a multi-class approach and a F1-Score of 0.89 for the COVID-19 identification in the hierarchical classification scenario. Due to the novelty of the application domain, there are still few works in the literature regarding the COVID-19 identification on CXR images.
READ FULL TEXT
