Research Article
Cardiac Magnetic Resonance Images Superresolution via Multichannel Residual Attention Networks
Table 1
The feature size of the MCRN for CMR images SR.
| | Network components | Kernel size | Stride | Padding | Input size | Output size |
| | Initial layer | 33 | 1 | 1 | HW1 | HW64 |
| | RAB | | | | | | | Conv 1 | 33 | 1 | 1 | HW64 | HW64 | | Conv 2 | 11 | 1 | 0 | HW64 | HW4 | | Conv 3 | 11 | 1 | 0 | HW4 | HW64 |
| | Up-block | | | | | | | DeConv | 44 | 2 | 1 | HW64 | HW128 | | Conv | 44 | 2 | 1 | HW128 | HW64 | | DeConv | 44 | 2 | 1 | HW64 | HW128 |
| | Down-block | | | | | | | Conv | 44 | 2 | 1 | HW128 | HW64 | | DeConv | 44 | 2 | 1 | HW64 | HW128 | | Conv | 44 | 2 | 1 | HW128 | HW64 |
| | Middle layer | 33 | 1 | 1 | HW64 | HW64 |
| | Upsample | 33 | 1 | 1 | HW64 | 2H2 W64 |
| | Reconstruction layer | 33 | 1 | 1 | 2H2 W64 | 2H2 W1 |
|
|
