Table 4 Results from other articles on Dataset-A. Results from the proposed CircWaveNet are shown in bold.
Number | Paper | Dimension | Method | ACC (%) | SE (%) | PR (%) | ROAUC (%) |
|---|---|---|---|---|---|---|---|
1 | Rasti et al.29 | 3D | Multi-scale Convolutional Mixture of Expert | – | – | 99.39 | 99.8 |
2 | Fang et al.54 | 3D | Lesion-Aware CNN | – | 99.36 | 99.39 | 99.80 |
3 | Das et al.55 | 3D | B-scan Attentive CNN | 93.2 | – | – | 95 |
4 | Rasti et al.56 | 3D | Wavelet-based Convolutional Mixture of Experts | – | – | – | 99.3 |
5 | Wang et al.57 | 3D | Volumetric OCT-Recurrent Neural Network | 93.8 | 94.0 | 94.4 | – |
6 | Wang et al.44 | 2D | CliqueNet | 98.6 | – | – | – |
7 | Das et al.58 | 2D | semi-supervised Generative Adversarial Network | 97.43 | 97.43 | – | – |
8 | Xu et al.59 | 2D | Multi-branch Hybrid Attention Network | 99.7 | – | 1 | – |
9 | Nabijiang et al.60 | 2D | Block Attention Mechanism | 99.64 | – | – | |
10 | Our Method | 2D | CircWaveNet | 94.5 | 96 | 90 | 98 |
