Table 3.
Metrics | Area Error Metrics | Boundary Error Metrics | Time | ||||||
---|---|---|---|---|---|---|---|---|---|
Methods | Ave. TPR | Ave. FPR | Ave. JI | Ave. DSC | Ave. AER | Ave. HE | Ave. MAE | Ave. Time (s) | |
FCN-AlexNet [60] | 0.95/-- | 0.34/-- | 0.74/-- | 0.84/-- | 0.39/-- | 25.1/-- | 7.1/-- | 5.8 | |
SegNet [61] | 0.94/-- | 0.16/-- | 0.82/-- | 0.89/-- | 0.22/-- | 21.7/-- | 4.5/-- | 12.1 | |
U-Net [55] | 0.92/-- | 0.14/-- | 0.83/-- | 0.90/-- | 0.22/-- | 26.8/-- | 4.9/-- | 2.15 | |
CE-Net [66] | 0.91/-- | 0.13/-- | 0.83/-- | 0.90/-- | 0.22/-- | 21.6/-- | 4.5/-- | 2.0 | |
MultiResUNet [65] | 0.93/-- | 0.11/-- | 0.84/-- | 0.91/-- | 0.19/-- | 18.8/-- | 4.1/-- | 6.5 | |
RDAU NET [63] | 0.91/-- | 0.11/-- | 0.84/-- | 0.91/-- | 0.19/-- | 19.3/-- | 4.1/-- | 3.5 | |
SCAN [64] | 0.91/-- | 0.11/-- | 0.83/-- | 0.90/-- | 0.20/-- | 26.9/-- | 4.9/-- | 4.1 | |
DenseU-Net [67] | 0.91/-- | 0.16/-- | 0.81/-- | 0.88/-- | 0.25/-- | 25.3/-- | 5.5/-- | 3.5 | |
STAN [21] | 0.92/-- | 0.09/-- | 0.85/-- | 0.91/-- | 0.18/-- | 18.9/-- | 3.9/-- | 5.8 | |
Xian, et al. [5] | 0.81/0.91 | 0.16/0.10 | 0.72/0.84 | 0.83/-- | 0.36/-- | 49.2/24.4 | 12.7/5.8 | 3.5 | |
Shan, et al. [15] | 0.81/0.93 | 1.06/0.13 | 0.60/-- | 0.70/-- | 1.25/-- | 107.6/18.9 | 26.6/5.0 | 3.0 | |
Shao, et al. [6] | 0.67/0.81 | 0.18/0.12 | 0.61/0.74 | 0.71/-- | 0.51/-- | 69.2/50.2 | 21.3/13.4 | 3.5 | |
Fuzzy FCN [62] | 0.94/-- | 0.08/-- | 0.88/-- | 0.92/-- | 0.14/-- | 19.8/-- | 4.2/-- | 6.0 | |
Huang, et al. [19] | 0.93/0.93 | 0.07/0.07 | 0.87/0.87 | 0.93/0.93 | 0.15/0.15 | 26.0/26.0 | 4.9/4.9 | 6.5 | |
Liu, et al. [4] LR = 1.5 |
0.82/0.94 | 0.13/0.08 | 0.73/0.87 | 0.84/-- | 0.31/-- | 44.0/26.3 | 10.4/-- | 27.0 | |
Liu, et al. [22] LR = 1.9 |
0.84/0.94 | 0.07/0.07 | 0.79/0.88 | 0.88/-- | 0.23/-- | 29.0/25.1 | 7.6/-- | 336.0 |
The values before the slashes are approaches’ performances on the proposed dataset, and after the slashes are their performances reported in the original publications. Notation ‘--’ indicates that the corresponding metric was not reported in the original paper. The best performance in each column is highlighted in bold.