Skip to main content
. 2022 Jun 3;16:902528. doi: 10.3389/fnins.2022.902528

Table 3.

Classification results for MCI-c vs. MCI-nc with different kinds of feature representations.

Feature representation OFMROI OFMnetwork CFM ROI CFM network
Parameters setting E1 = 2 E1 = 3 E1 = 4 E1 = 2 E1 = 3 E1 = 4
E2 = 1 E2 = 2 E2 = 3 E2 = 4 E2 = 5 E2 = 1 E2 = 2 E2 = 3 E2 = 4 E2 = 5 E2 = 1 E2 = 2 E2 = 3 E2 = 4 E2 = 5
ACC (%) 64.15 68.62 75.10 73.81 71.31 70.24 72.38 78.92 76.07 73.25 80.75 87.72 82.31 74.59 71.76 71.22 78.63 79.41 75.71 70.65
SEN (%) 62.30 66.44 72.23 70.54 67.90 66.21 69.70 75.92 73.82 70.61 78.42 84.16 78.73 72.16 69.02 68.78 76.23 77.20 71.02 68.99
SPE (%) 66.74 71.46 79.42 76.42 73.54 72.44 76.07 81.13 79.16 77.25 84.57 91.64 84.79 79.85 75.59 75.28 81.78 82.18 78.29 73.06

OFMROI, the original ROI-level feature representation; OFMnetwork, the original network-level feature representation; CFMROI, the coupled ROI-level feature representation; CFMnetwork, the two-level feature representation; Acc, accuracy; Sen, sensitivity; Spe, specificity. The values with the highest accuracy are highlighted in boldface. The shadow of gray color is used to visually differentiate columns of the table.