From: CRANet: a comprehensive residual attention network for intracranial aneurysm image classification
Models | 2-classification | 4-classification | ||||
---|---|---|---|---|---|---|
 | Accuracy | Recall | F1 | Accuracy | Recall | F1 |
ResNet18 | 96.57 | 0.91 | 0.92 | 91.75 | 0.87 | 0.88 |
ResNet34 | 96.10 | 0.70 | 0.73 | 90.25 | 0.88 | 0.87 |
ResNet50 | 95.64 | 0.84 | 0.86 | 89.15 | 0.85 | 0.85 |
ResNet101 | 95.23 | 0.89 | 0.91 | 89.60 | 0.85 | 0.84 |
VGG | 96.64 | 0.89 | 0.88 | 90.85 | 0.85 | 0.86 |
GoogleNet | 96.87 | 0.89 | 0.89 | 86.50 | 0.81 | 0.79 |
InceptionV3 | 97.16 | 0.91 | 0.92 | 91.68 | 0.87 | 0.86 |
DenseNet | 96.32 | 0.89 | 0.87 | 88.79 | 0.84 | 0.84 |
CNN | 95.89 | 0.87 | 0.85 | 86.27 | 0.82 | 0.81 |
CRANet (Our model) | 97.81 | 0.94 | 0.94 | 92.55 | 0.91 | 0.91 |