Table 3 Performance comparisons of the proposed method against state-of-the-art methods on DRIVE dataset using different performance metrics
From: DNL-Net: deformed non-local neural network for blood vessel segmentation
Method | Sen (%) | Acc (%) | AUC (%) |
|---|---|---|---|
Azzopadi et al. [53] | 76.55 | 94.42 | 96.14 |
Roychowdhury et al. [54] | 72.50 | 95.20 | 96.72 |
Zhao et al. [55] | 74.20 | 95.40 | 86.20 |
U-Net [15] | 73.44 | 95.23 | 97.44 |
DeepVessel [56] | 76.03 | 95.23 | 97.52 |
Li et al. [57] | 75.69 | 95.27 | 97.38 |
Melinscak et al. [58] | – | 94.66 | 97.49 |
DEU-Net [16] | 79.40 | 95.67 | 97.72 |
CE-Net [17] | 83.09 | 95.45 | 97.79 |
DenseU-Net [7] | 80.40 | 96.04 | 97.97 |
R2U-Net [36] | 83.18 | 95.93 | 98.11 |
SA-Net [42] | 82.52 | 95.69 | 98.22 |
SSCA-Net [35] | 83.52 | 96.14 | 98.20 |
DNL-Net | 86.64 | 96.10 | 98.37 |