Table 6 Comparison of DICE for image rotation and resizing.
From: CFM-UNet: coupling local and global feature extraction networks for medical image segmentation
Image rotation | Pixel resizing | ||||||||
|---|---|---|---|---|---|---|---|---|---|
\(0^\circ\) | \(60^\circ\) | \(120^\circ\) | \(180^\circ\) | \(240^\circ\) | \(300^\circ\) | \(256\times 256\) | \(512\times 512\) | \(1024\times 1024\) | |
DICE | 95.41% | 91.15% | 88.72% | 87.46% | 87.18% | 91.70% | 79.80% | 95.41% | 95.40% |
TPR | 96.01% | 92.26% | 89.86% | 88.83% | 87.31% | 94.03% | 90.83% | 96.01 | 95.95% |
VOE | 8.52% | 15.65% | 18.69% | 20.05% | 20.66% | 14.83% | 31.07% | 8.52% | 8.55% |
RVD | 0.70% | 7.42% | 0.13% | 2.40% | 0.91% | 3.56% | 20.23% | 0.70% | 0.59% |
ASSD | 0.95% | 1.85% | 2.11% | 2.33% | 2.38% | 1.69% | 3.66% | 0.95% | 0.96% |
