Table 5 Results of comparison experiments on the COVID19-CT dataset.
From: Multi-branch CNN and grouping cascade attention for medical image classification
Method(year) | Params (M) | FLOPs (G) | Acc | F1 | Precision | Recall | Auc |
|---|---|---|---|---|---|---|---|
ResNet50 (2016) | 23.5 | 4.1 | 0.8716 | 0.8774 | 0.8947 | 0.8608 | 0.8724 |
MobileNetV2 (2018) | 2.2 | 0.3 | 0.8716 | 0.8805 | 0.8750 | 0.8861 | 0.8706 |
EfficientNet-B0 (2019) | 4.0 | 0.4 | 0.9054 | 0.9079 | 0.9452 | 0.8734 | 0.9077 |
RepVGG (2021) | 43.7 | 9.9 | 0.8986 | 0.9057 | 0.9000 | 0.9114 | 0.8977 |
ConvNext-S (2022) | 49.5 | 8.7 | 0.7838 | 0.7922 | 0.8133 | 0.7722 | 0.7846 |
ConvMixer (2023) | 47.9 | 49.1 | 0.8851 | 0.8957 | 0.8690 | 0.9241 | 0.8823 |
InceptionNext-S (2023) | 47.1 | 8.4 | 0.8581 | 0.8662 | 0.8718 | 0.8608 | 0.8579 |
FasterNet (2023) | 13.7 | 1.9 | 0.8784 | 0.8875 | 0.8765 | 0.8987 | 0.8769 |
Swin-S (2021) | 48.8 | 8.6 | 0.7027 | 0.7609 | 0.6667 | 0.8861 | 0.6894 |
CrossViT 18 (2021) | 43.3 | 9.0 | 0.7703 | 0.8023 | 0.7419 | 0.8734 | 0.7628 |
MoCoViT 1.0 (2022) | 7.2 | 0.5 | 0.8784 | 0.8889 | 0.8675 | 0.9114 | 0.8760 |
BiFormer-S (2023) | 56.0 | 9.4 | 0.8851 | 0.8903 | 0.9079 | 0.8734 | 0.8860 |
FasterViT-2 (2023) | 75.2 | 8.9 | 0.8716 | 0.8742 | 0.9167 | 0.8354 | 0.8742 |
Flatten-pvt (2023) | 24.2 | 3.7 | 0.8581 | 0.8591 | 0.9143 | 0.8101 | 0.8616 |
T ransxNet (2023) | 25.5 | 4.6 | 0.8649 | 0.8701 | 0.8933 | 0.8481 | 0.8661 |
GroupMixFormer (2023) | 22.1 | 5.1 | 0.9054 | 0.9103 | 0.9221 | 0.8987 | 0.9059 |
Eff-CTNet(Ours) | 25.2 | 6.4 | 0.9257 | 0.9317 | 0.9146 | 0.9494 | 0.9240 |
