Table 1 Segmentation performance comparison between different network architecture settings
From: RU-Net: skull stripping in rat brain MR images after ischemic stroke with rat U-Net
Network architecture | Evaluation metric | ||
|---|---|---|---|
ΚD(%) | Κst(%) | Κsb(%) | |
Maximum pooling:\(3\times 3\) | 95.20 ± 2.53 | 96.32 ± 3.23 | 93.77 ± 4.36 |
Maximum pooling:\(4\times 4\) | 95.18 ± 2.61 | 96.24 ± 3.29 | 93.82 ± 4.55 |
Convolution filter:\(7\times 7\) | 95.96 ± 2.42 | 96.92 ± 2.97 | 94.88 ± 2.98 |
U-shape structure depth: 4 | 98.01 ± 1.50 | 97.78 ± 1.85 | 98.06 ± 1.87 |
RU-Net | 98.04 ± 0.33 | 97.94 ± 0.75 | 98.15 ± 0.68 |