“…Several methods for lung segmentation have been proposed over the last two decades. Conventional methods rely on techniques such as thresholding [27,28], region growing [29,30], active contours [31,32], mathematical morphology [33,34], and cluster analysis [35][36][37]; however, deep learning (DL) approaches, in particular convolutional neural networks [38,39], generative adversarial networks [40,41], and residual neural networks [42], have recently gained popularity in this field as well. While DL methods achieve state-ofthe-art accuracy, a drawback of these methods is that they require substantial amounts of annotated training data to achieve the desired accuracy.…”