Table 12 Evaluation across imaging variations.
Model | Quality (PPI) | Resolution | Number of instances | Accuracy | Sensitivity | Specificity | Precision | Recall | AUC |
|---|---|---|---|---|---|---|---|---|---|
CNN-VGG | Low | < 150 | 120 | 75.368 | 74.365 | 76.426 | 75.635 | 76.576 | 0.768 |
Medium | 150–300 | 500 | 82.345 | 85.426 | 84.254 | 85.982 | 85.324 | 0.856 | |
High | > 300 | 2000 | 95.375 | 94.987 | 95.124 | 94.689 | 95.361 | 0.952 | |
CNN-GoogleNet | Low | < 150 | 120 | 77.821 | 78.956 | 77.985 | 78.451 | 77.961 | 0.783 |
Medium | 150–300 | 500 | 81.542 | 82.347 | 83.246 | 84.369 | 85.347 | 0.845 | |
High | > 300 | 2000 | 95.385 | 96.214 | 96.245 | 95.783 | 95.237 | 0.957 | |
CNN-DenseNet | Low | < 150 | 120 | 79.526 | 78.954 | 78.628 | 77.987 | 78.234 | 0.789 |
Medium | 150–300 | 500 | 85.627 | 85.397 | 86.354 | 87.124 | 86.872 | 0.862 | |
High | > 300 | 2000 | 95.846 | 96.247 | 96.328 | 96.429 | 95.876 | 0.964 |
