Skip to main content
Fig. 4 | BMC Medical Imaging

Fig. 4

From: Three-stage segmentation of lung region from CT images using deep neural networks

Fig. 4

Binary image tagged as a containing and b not containing lung region in the CNN-1 training data. Connected component tagged as c containing and d not containing lung region in the CNN-2 training data. e Binary mask of a grayscale CT slice that doubles as training image and training label in UNET-1 training data. f Plot of loss versus number of iterations during the training of UNET-1. g Zoomed-in version of the plot in the preceding subfigure. h CT slice from the ILD database. i Binary image of the slice in the preceding subfigure. j Lung region of the CT slice from the ILD database. k Contour image of the lung region shown in the preceding subfigure. l Plot of loss versus number of iterations during the training of UNET-2

Back to article page