Table 2 Accuracy (ACC) and AUC score of models based on features selected by rMLTFL and ANOVA (p value threshold=0.05) respectively
Sinkhorn distance | Sinkhorn distance + L1l2 | Sinkhorn distance + Lpl1 | |||||
|---|---|---|---|---|---|---|---|
ACC | AUC | ACC | AUC | ACC | AUC | ||
FDG | rMLTFL | 68.76 ±7.53 | 0.66 ±0.08 | 66.04 ±7.53 | 0.65 ±0.08 | 65.48 ±5.04 | 0.64 ±0.07 |
ANOVA | 66.07 ±6.96 | 0.65 ±0.07 | 63.63 ±6.01 | 0.64 ±0.07 | 59.50 ±6.53 | 0.62 ±0.08 | |
VBM | rMLTFL | 62.37 ±6.88 | 0.62 ±0.11 | 62.74 ±0.08 | 0.62 ±0.11 | 57.86 ±6.32 | 0.60 ±0.07 |
ANOVA | 58.94 ±7.82 | 0.59 ±0.12 | 58.68 ±0.08 | 0.58 ±0.12 | 56.79 ±0.11 | 0.58 ±0.13 | |
FDG + VBM | rMLTFL | 62.26 ±6.48 | 0.63 ±0.05 | 66.61 ±6.29 | 0.65 ±0.06 | 66.05 ±5.91 | 0.65 ±0.08 |
ANOVA | 61.44 ±6.23 | 0.64 ±0.05 | 63.87 ±5.75 | 0.63 ±0.06 | 61.15 ±8.01 | 0.64 ±0.07 | |