Table 3.

5-fold cross-validation accuracy for the classification of ADNI participants, for ModelA ModelB ModelC and ModelD with all numbers of channels c, and the best Linear SVM. The number of channels corresponds to the number of 3D convolutional kernels used in each CNN convolutional layer.

c	ModelA	ModelB	ModelC	ModelD
24	76.41%	82.65%	75.47%	58.59%
28	83.85%	84.45%	65.59%	54.78%
32	78.86%	85.44%	84.45%	58.20%
36	83.84%	86.05%	72.18%	55.80%
40	78.05%	86.25%	58.18%	47.21%
44	61.80%	87.25%	55.65%	52.99%
48	59.81%	86.45%	64.45%	55.48%
52	61.41%	86.46%	62.64%	47.81%

Best SVM-C parameter 0.001, accuracy 81.19%.