Table I.

Classification performance of the models developed without the ML for our dataset. The standard deviations of the parameters were calculated using bootstrap.

Method	AUC	Accuracy	Sensitivity	Specificity
Fine tuning	0.895±0.031	0.860±0.024	0.848±0.056	0.863±0.026
MP features	0.849±0.036	0.793±0.036	0.757±0.054	0.803±0.048
FC features	0.849±0.036	0.800±0.038	0.757±0.054	0.812±0.046
Fine tuning, ML	0.936±0.019	0.887±0.028	0.848±0.039	0.897±0.035
MP features, ML	0.889±0.029	0.860±0.044	0.757±0.058	0.889±0.062
FC features, ML	0.873±0.036	0.753±0.044	0.879±0.058	0.718±0.053