Table 10 Quantitative evaluation (RMSE, SNR, PSNR, and SSIM) of different up-sampling factors using BrainWeb MR images
From: Fusing multi-scale information in convolution network for MR image super-resolution reconstruction
Eval. met | Scale | Bicubic | NLM | Sparse coding | SRCNN | MFCN |
|---|---|---|---|---|---|---|
RMSE | 2 | 2.5077 | 2.1112 | 1.7747 | 1.4836 | 1.2026 |
3 | 4.2038 | 3.7707 | 3.4262 | 2.8937 | 2.5415 | |
4 | 5.849 | 5.5588 | 5.0577 | 4.6946 | 4.5196 | |
SNR | 2 | 27.1376 | 28.636 | 30.1495 | 31.7244 | 33.5375 |
3 | 22.6394 | 23.5894 | 24.422 | 25.9595 | 27.0741 | |
4 | 19.767 | 20.2123 | 21.0326 | 21.7213 | 22.0571 | |
PSNR | 2 | 40.1699 | 41.6684 | 43.1819 | 44.7567 | 46.5698 |
3 | 35.6717 | 36.6218 | 37.4262 | 38.9918 | 40.1064 | |
4 | 32.7993 | 33.2446 | 34.056 | 34.7536 | 35.0895 | |
SSIM | 2 | 0.9891 | 0.9923 | 0.9945 | 0.9963 | 0.9975 |
3 | 0.9678 | 0.9743 | 0.9788 | 0.9864 | 0.9891 | |
4 | 0.9375 | 0.9434 | 0.9529 | 0.9639 | 0.9662 |