Figure 2
From: Deep convolutional neural networks for automated scoring of pentagon copying test results
ROC curve in the validation dataset for prediction of the PCT images being categorized as correct based on P(PCTcorrect) values calculated by the fine-tuned GoogLeNet CNN. aThe cut-off probability (specificity, sensitivity) is shown at the point closest to the top left-hand corner. CNN, convolutional neural network; PCT, pentagon copying test; P(PCTcorrect), CNN-calculated probability of the PCT image being categorized as correct; ROC, receiver operating characteristic.
