Table 2.
The results of all combinations (Model_REZ) in internal test set.
Models | AUC (95%CI) | Sen (%) | Spe (%) | Acc (%) |
---|---|---|---|---|
Xception | ||||
224×224 | 0.896 (0.880-0.913) | 77.31 | 87.74 | 83.53 |
320×320 | 0.883 (0.866-0.899) | 81.22 | 81.08 | 81.14 |
448×448 | 0.887 (0.869-0.904) | 83.41 | 82.14 | 82.65 |
MobileNet | ||||
224×224 | 0.877 (0.859-0.895) | 80.75 | 81.40 | 81.14 |
320×320 | 0.867 (0.849-0.886) | 81.38 | 78.54 | 79.68 |
448×448 | 0.886 (0.870-0.903) | 80.59 | 80.97 | 80.59 |
EfficientNetB0 | ||||
224×224 | 0.878 (0.861-0.895) | 79.34 | 81.92 | 80.88 |
320×320 | 0.887 (0.870-0.904) | 80.28 | 83.30 | 82.08 |
448×448 | 0.875 (0.857-0.893) | 74.49 | 89.01 | 83.15 |
ResNet50 | ||||
224×224 | 0.781 (0.758-0.804) | 72.46 | 70.08 | 71.04 |
320×320 | 0.847 (0.827-0.866) | 77.93 | 76.96 | 77.35 |
448×448 | 0.851 (0.832-0.870) | 72.61 | 82.88 | 78.74 |
DenseNet121 | ||||
224×224 | 0.867 (0.849-0.885) | 77.46 | 81.40 | 79.81 |
320×320 | 0.849 (0.829-0.869) | 79.34 | 78.54 | 78.86 |
448×448 | 0.866 (0.848-0.884) | 79.03 | 80.97 | 80.19 |
REZ, resolution; AUC, area under the curve; CI, confidence interval; Sen, sensitivity; Spe, specificity; Acc, accuracy.