Skip to main content
. 2020 Nov 27;51(5):2864–2889. doi: 10.1007/s10489-020-02010-w

Table 3.

Results obtained from pre-trained CNN models for Binary Class Dataset1

Models/Folds(F) Performance Evaluation based on four metrics
TP TN FP FN Test set (%) Acc Rec Spec Prec F1
AlexNet F-1 42 92 9 3 91.78 93.33 91.09 82.35 87.5
F-2 44 98 3 1 97.26 97.78 97.03 93.62 95.65
F-3 45 98 3 0 97.95 100 97.03 93.75 96.77
F-4 45 98 3 0 97.95 100 97.03 93.75 96.77
F-5 45 98 3 0 97.95 100 97.03 93.75 96.77
Average 96.58 98.22 95.84 91.44 94.69
GoogleNet F-1 42 93 8 3 92.47 93.33 92.08 84 88.42
F-2 45 97 4 0 97.26 100 96.04 91.84 95.74
F-3 45 99 2 0 98.63 100 98.02 95.74 97.83
F-4 45 99 2 0 98.63 100 98.02 95.74 97.83
F-5 45 99 2 0 98.63 100 98.02 95.74 97.83
Average 97.12 98.67 96.44 92.61 95.53
DenseNet121 F-1 43 91 10 2 91.78 95.56 90.1 81.13 87.76
F-2 44 97 4 1 96.58 97.78 96.04 91.67 94.62
F-3 44 100 1 1 98.63 97.78 99.01 97.78 97.78
F-4 44 100 1 1 98.63 97.78 99.01 97.78 97.78
F-5 44 100 1 1 98.63 97.78 99.01 97.78 97.78
Average 96.85 97.33 96.63 93.23 95.14
ResNet-50 F-1 43 93 8 2 93.15 95.56 92.08 84.31 89.58
F-2 45 98 3 0 97.95 100 97.03 93.75 96.77
F-3 45 99 2 0 98.63 100 98.02 95.74 97.83
F-4 45 99 2 0 98.63 100 98.02 95.74 97.83
F-5 45 99 2 0 98.63 100 98.02 95.74 97.83
Average 97.4 99.11 96.63 93.06 95.97
Se-ResNet 50 F-1 43 94 7 2 93.84 95.56 93.07 86 90.53
F-2 44 99 2 1 97.95 97.78 98.02 95.65 96.7
F-3 45 99 2 0 98.63 100 98.02 95.74 97.83
F-4 45 99 2 0 98.63 100 98.02 95.74 97.83
F-5 45 99 2 0 98.63 100 98.02 95.74 97.83
Average 97.53 98.67 97.03 93.78 96.14
Inception ResNet v2 F-1 41 87 14 4 87.67 91.11 86.14 74.55 82
F-2 43 90 11 2 91.1 95.56 89.11 79.63 86.87
F-3 43 90 11 2 91.1 95.56 89.11 79.63 86.87
F-4 41 88 13 4 88.36 91.11 87.13 75.93 82.83
F-5 41 89 12 4 89.04 91.11 88.12 77.36 83.67
Average 89.45 92.89 87.92 77.42 84.45
ResNeXt-50 F-1 43 95 6 2 94.52 95.56 94.06 87.76 91.49
F-2 45 98 3 0 97.95 100 97.03 93.75 96.77
F-3 45 100 1 0 99.32 100 99.01 97.83 98.9
F-4 45 100 1 0 99.32 100 99.01 97.83 98.9
F-5 45 100 1 0 99.32 100 99.01 97.83 98.9
Average 98.08 99.11 97.62 95 96.99
Se-ResNeXt-50 F-1 43 96 5 2 95.21 95.56 95.05 89.58 92.47
F-2 45 99 2 0 98.63 100 98.02 95.74 97.83
F-3 45 100 1 0 99.32 100 99.01 97.83 98.9
F-4 45 100 1 0 99.32 100 99.01 97.83 98.9
F-5 45 100 1 0 99.32 100 99.01 97.83 98.9
Average 98.36 99.11 98.02 95.76 97.4

Bold indicates highest accuracy