Table 3 Target detection sensitivities (including non-visible class) for the eight studied CNNs
From: Convolutional neural network -based phantom image scoring for mammography quality control
Sensitivity (%) | ||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
f1 | f2 | f3 | f4 | f5 | f6 | c1 | c2 | c3 | c4 | c5 | m1 | m2 | m3 | m4 | m5 | nonv | Acc | |
CNN1 | 100.0 | 95.9 | 95.1 | 92.7 | 94.8 | 88.0 | 99.4 | 95.6 | 91.2 | 94.9 | 84.2 | 100.0 | 94.9 | 83.8 | 90.8 | 96.7 | 78.3 | 91.0 |
CNN2 | 98.7 | 96.6 | 94.4 | 96.0 | 93.8 | 76.0 | 100.0 | 97.5 | 93.9 | 95.7 | 84.2 | 100.0 | 98.1 | 89.6 | 97.9 | 92.2 | 85.4 | 93.6 |
CNN3 | 100.0 | 98.0 | 97.2 | 95.2 | 91.7 | 72.0 | 100.0 | 98.1 | 96.6 | 97.4 | 89.5 | 100.0 | 97.4 | 96.1 | 99.3 | 96.7 | 85.4 | 94.6 |
CNN4 | 100.0 | 98.6 | 98.6 | 96.0 | 93.8 | 80.0 | 100.0 | 98.7 | 93.9 | 98.3 | 89.5 | 100.0 | 98.7 | 96.1 | 97.2 | 94.4 | 87.6 | 95.2 |
CNN5 | 99.4 | 98.6 | 97.2 | 94.4 | 93.8 | 88.0 | 100.0 | 100.0 | 97.3 | 97.4 | 89.5 | 100.0 | 98.7 | 96.1 | 98.6 | 95.6 | 85.4 | 95.0 |
CNN6 | 100.0 | 98.0 | 95.8 | 96.0 | 94.8 | 80.0 | 100.0 | 99.4 | 93.9 | 96.6 | 68.4 | 100.0 | 98.7 | 97.4 | 97.9 | 97.8 | 84.4 | 94.4 |
CNN7 | 100.0 | 97.3 | 96.5 | 93.5 | 95.8 | 92.0 | 100.0 | 98.7 | 94.6 | 94.9 | 68.4 | 100.0 | 99.4 | 96.8 | 98.6 | 96.7 | 83.8 | 94.3 |
CNN8 | 100.0 | 99.3 | 98.6 | 96.0 | 97.9 | 88.0 | 100.0 | 99.4 | 96.6 | 94.9 | 73.7 | 100.0 | 98.7 | 95.5 | 97.2 | 94.4 | 82.2 | 94.2 |