Skip to main content
. 2021 Apr 9;15:646013. doi: 10.3389/fnins.2021.646013

TABLE 4.

Results (%) of the models trained from only PET data for CN vs. AD task.

Model ACC SEN SPE AUC
3D-VGG11 (Huang et al., 2019) 89.11 90.24 87.77 92.69
3D-ResNet10 86.26 86.56 85.94 84.48
3D CNN (4 layers) 88.77 89.11 88.41 87.11

Bold values mean the best value in the comparative experiment.