研究文章
三维脑ML图像并行遗留学习使用全球局部特征提取
表7
PSNR(顶部)和SSIM(底部)对不同算法和不同空间分辨率的比较
|
| 空间分辨率
) |
方法论 |
| BM3D |
CNN-DMRI |
里钱网 |
3D-Parallel-RicianNet |
|
|
|
31.8908 0.9709 |
34.7330 0.9890 |
35.7466 0.9233 |
41.5169 0.9970 |
|
|
32.1794 0.9718 |
35.2350 0.9903 |
36.4030 0.9117 |
41.9756 0.9973 |
|
|
33.1937 0.9725 |
35.4720 0.9891 |
70.0095 0.806 |
9950 0.9982 |
|
|
317986 0.9677 |
3606 0.9917 |
36.6172 0.9078 |
42.2193 0.9976 |
|
|
31.3591 0.963 |
34.0402 0.9917 |
36.0542 0.9070 |
41.9259 0.9974 |
|
|
298656 0.948 |
351568 0.9918 |
31.4012 0.9180 |
40.3297 0.9961 |
|
|
28.2362 0.9230 |
34.5695 0.9900 |
23.4731 0.9038 |
38.4990 0.9933 |
|
|
