Table 2.

Performances of the five variants of the AD-classification network.

Input	Model	Sensitivity	Specificity	ROC AUC	Accuracy @ Operating	Max accuracy
T1W	Regular	0.885	0.860	0.905	0.869	0.873
SynCBV	Regular	0.885	0.876	0.919	0.877	0.881
T1W+SynCBV	Dual-channel	0.885	0.853	0.936	0.865	0.869
T1W+SynCBV	Dual-encoder w/ identical weights	0.802	0.806	0.875	0.800	0.804
T1W+SynCBV	Dual-encoder w/ different weights	0.901	0.876	0.942	0.885	0.888

Sensitivity and specificity are calculated at the operating point. Accuracy at the operating point and the maximum accuracy achievable by changing the binarization threshold are respectively calculated for each candidate. ROC AUC, area under the receiver-operating characteristics curve; SynCBV, synthesized CBV. Best result(s) in each metric are highlighted in bold.