Impact of lung segmentation on the diagnosis and explanation of COVID-19 in chest X-ray images

COVID-19 frequently provokes pneumonia, which can be diagnosed using imaging\nexams. Chest X-ray (CXR) is often useful because it is cheap, fast, widespread,\nand uses less radiation. Here, we demonstrate the impact of lung segmentation\nin COVID-19 identification using CXR images and evaluate which contents of the\nimage influenced the most. Semantic segmentation was performed using a U-Net\nCNN architecture, and the classification using three CNN architectures (VGG,\nResNet, and Inception). Explainable Artificial Intelligence techniques were\nemployed to estimate the impact of segmentation. A three-classes database was\ncomposed: lung opacity (pneumonia), COVID-19, and normal. We assessed the\nimpact of creating a CXR image database from different sources, and the\nCOVID-19 generalization from one source to another. The segmentation achieved a\nJaccard distance of 0.034 and a Dice coefficient of 0.982. The classification\nusing segmented images achieved an F1-Score of 0.88 for the multi-class setup,\nand 0.83 for COVID-19 identification. In the cross-dataset scenario, we\nobtained an F1-Score of 0.74 and an area under the ROC curve of 0.9 for\nCOVID-19 identification using segmented images. Experiments support the\nconclusion that even after segmentation, there is a strong bias introduced by\nunderlying factors from different sources.\n