Table 9 Summary of results from state-of-the-art pre-trained ViT models.
From: Computer-aided cholelithiasis diagnosis using explainable convolutional neural network
Model | Model size | Accuracy | Precision | Recall | Specificity | F1-score | AUC score |
|---|---|---|---|---|---|---|---|
ViT Base- 16 | 330 MB | 0.80 | 0.70 | 0.85 | 0.83 | 0.72 | 0.81 |
ViT Base- 32 | 337 MB | 0.87 | 0.75 | 0.84 | 0.80 | 0.78 | 0.80 |
ViT Large- 16 | 1.13 GB | 0.77 | 0.69 | 0.84 | 0.89 | 0.70 | 0.89 |
ViT Large- 32 | 1.14 GB | 0.76 | 0.67 | 0.81 | 0.63 | 0.68 | 0.79 |
SWIN transformer | 108 MB | 0.77 | 0.67 | 0.80 | 0.80 | 0.68 | 0.74 |
Proposed custom CNN | 355 KB | 0.88 | 0.90 | 0.85 | 0.89 | 0.87 | 0.87 |
