Table 3.
Summary of the models trained from the Mask, MaskedImage, and ImageOnly groups for CN vs. AD task.
| MRI ROI | ACC1 | SEN | SPE | AUC |
|---|---|---|---|---|
| Mask | 76.57% | 83.87% | 71.51% | 84.24% |
| Maskedlmage | 79.21% | 76.61% | 81.01% | 84.63% |
| ImageOnly | 84.82% | 87.90% | 82.68% | 87.47% |
The Segmented dataset was used. 1ACC, SEN, SPE, AUC denotes accuracy, sensitivity, specificity and area under curve, respectively. When testing, the numbers of true positive (TP), true negative (TN), false negative (FP), and false negative (FN) subjects were counted, as ACC = (TP+TN)/(TP+TN+FP+FN), SEN = TP/(TP+FN), SPE = TN/(TN+FP). AUC is obtained through calculating the area under the receiver operating characteristic (ROC) curve. For all four metrics, the values are between 0 and 100%, the higher, the better.