Skip to main content
. 2021 Apr 16;11(4):310. doi: 10.3390/jpm11040310

Figure 1.

Figure 1

(1) STL models are produced from 58 CT scans and then overlapped with the MR images to create the first dataset. (2) 62 MR images are used to create brain STL models, and a brain segmentation algorithm is created. The brain segmentation algorithm is combined with manual corrections to remove the brain from dataset 1 to create dataset 2. (3) Finally, these 2 datasets are compared using the same CNN topology.