Table 4.
Classification report of CNN models (data augmentation performed and obtained 1600 images, of which 1280 were training images and 320 were validation images).
| Number of Epochs = 5 | |||||
|---|---|---|---|---|---|
| CNN Model | Grade | Precision | Recall | F1-Score | Accuracy |
| Model 1: EfficientNetV2L |
Grade 0 | 0.59 | 0.90 | 0.71 | 0.70 |
| Grade 1 | 0.92 | 0.50 | 0.64 | ||
| Grade 2 | 0.68 | 0.80 | 0.73 | ||
| Model 2: ResNet152V2 |
Grade 0 | 0.31 | 0.90 | 0.47 | 0.37 |
| Grade 1 | 0.49 | 0.29 | 0.36 | ||
| Grade 2 | 0.55 | 0.06 | 0.10 | ||
| Model 3: DenseNet201 |
IDC Grade 0 | 0.38 | 0.82 | 0.52 | 0.44 |
| Grade 1 | 0.45 | 0.10 | 0.16 | ||
| Grade 2 | 0.53 | 0.60 | 0.56 | ||
| Ensemble of CNN model (Model 1 + Model 2 + Model 3) | Grade 0 | 0.78 | 0.72 | 0.75 | 0.79 |
| Grade 1 | 0.78 | 0.83 | 0.81 | ||
| Grade 2 | 0.82 | 0.80 | 0.81 | ||
| Number of epochs = 10 | |||||
| CNN model | Grade | Precision | Recall | F1-Score | Accuracy |
| Model 1: EfficientNetV2L |
Grade 0 | 0.27 | 0.03 | 0.06 | 0.32 |
| Grade 1 | 0.39 | 0.09 | 0.15 | ||
| Grade 2 | 0.32 | 0.85 | 0.46 | ||
| Model 2: ResNet152V2 |
Grade 0 | 0.00 | 0.00 | 0.00 | 0.40 |
| Grade 1 | 0.40 | 1.00 | 0.57 | ||
| Grade 2 | 0.00 | 0.00 | 0.00 | ||
| Model 3: DenseNet201 |
Grade 0 | 0.00 | 0.00 | 0.00 | 0.38 |
| Grade 1 | 0.39 | 0.93 | 0.55 | ||
| Grade 2 | 0.22 | 0.02 | 0.04 | ||
| Ensemble of CNN model (Model 1 + Model 2 + Model 3) | Grade 0 | 0.79 | 0.89 | 0.83 | 0.86 |
| Grade 1 | 0.92 | 0.84 | 0.87 | ||
| Grade 2 | 0.86 | 0.86 | 0.86 | ||
| Number of epochs = 15 | |||||
| CNN model | Grade | Precision | Recall | F1-Score | Accuracy |
| Model 1: EfficientNetV2L |
Grade 0 | 0.36 | 0.04 | 0.07 | 0.34 |
| Grade 1 | 0.35 | 0.91 | 0.51 | ||
| Grade 2 | 0.21 | 0.05 | 0.09 | ||
| Model 2: ResNet152V2 |
Grade 0 | 0.32 | 1.00 | 0.48 | 0.32 |
| Grade 1 | 0.00 | 0.00. | 0.00 | ||
| Grade 2 | 0.00 | 0.00 | 0.00 | ||
| Model 3: DenseNet201 |
Grade 0 | 0.26 | 0.56 | 0.36 | 0.25 |
| Grade 1 | 0.26 | 0.22 | 0.24 | ||
| Grade 2 | 0.00 | 0.00 | 0.00 | ||
| Ensemble of CNN model (Model 1 + Model 2 + Model 3) | Grade 0 | 0.92 | 0.81 | 0.86 | 0.86 |
| Grade 1 | 0.81 | 0.94 | 0.87 | ||
| Grade 2 | 0.88 | 0.83 | 0.85 | ||
| Number of epochs = 20 | |||||
| CNN model | Grade | Precision | Recall | F1-Score | Accuracy |
| Model 1: EfficientNetV2L |
Grade 0 | 0.24 | 0.64 | 0.35 | 0.24 |
| Grade 1 | 0.45 | 0.04 | 0.07 | ||
| Grade 2 | 0.19 | 0.15 | 0.16 | ||
| Model 2: ResNet152V2 |
Grade 0 | 0.23 | 0.14 | 0.17 | 0.39 |
| Grade 1 | 0.42 | 0.88 | 0.57 | ||
| Grade 2 | 0.00 | 0.00 | 0.00 | ||
| Model 3: DenseNet201 |
Grade 0 | 0.43 | 0.28 | 0.34 | 0.47 |
| Grade 1 | 0.48 | 0.67 | 0.56 | ||
| Grade 2 | 0.48 | 0.39 | 0.43 | ||
| Ensemble of CNN model (Model 1 + Model 2 + Model 3) | Grade 0 | 0.86 | 0.85 | 0.86 | 0.87 |
| Grade 1 | 0.85 | 0.89 | 0.87 | ||
| Grade 2 | 0.89 | 0.84 | 0.87 | ||