Longitudinal AV45-PET generation on ADNI 8
25.1723Whole-Image PSNR∆-Diffusion
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| ∆-DiffusionGPU=NVIDIA H100, Optimizer=Adam, Learning rate=1×10−4, Batch size=12026.06 | 25.1723 | 82.84 | 34.2089 | 98.06 | — | |
| SiM2PGPU=NVIDIA H100, Optimizer=Adam, Learning rate=1×10−4, Batch size=12026.06 | 24.5516 | 81.52 | 33.4698 | 97.94 | 3.18 | |
| VAEGANGPU=NVIDIA H100, Optimizer=Adam, Learning rate=1×10−4, Batch size=12026.06 | 23.958 | 81.48 | 33.3028 | 97.98 | 1.68 | |
| FICDGPU=NVIDIA H100, Optimizer=Adam, Learning rate=1×10−4, Batch size=12026.06 | 23.7418 | 81.35 | 32.6071 | 95.26 | 1.04 | |
| LDMGPU=NVIDIA H100, Optimizer=Adam, Learning rate=1×10−4, Batch size=12026.06 | 22.0798 | 79.5 | 31.09 | 97.56 | 5.29 | |
| DDPMGPU=NVIDIA H100, Optimizer=Adam, Learning rate=1×10−4, Batch size=12026.06 | 20.1783 | 75.47 | 31.1558 | 97.43 | 5.98 | |
| DDIMGPU=NVIDIA H100, Optimizer=Adam, Learning rate=1×10−4, Batch size=12026.06 | 19.0182 | 73.91 | 29.6995 | 97.18 | 1.37 | |
| ControlNetGPU=NVIDIA H100, Optimizer=Adam, Learning rate=1×10−4, Batch size=12026.06 | 18.3631 | 72.97 | 28.1682 | 96.52 | 5.83 | |
| PNDMGPU=NVIDIA H100, Optimizer=Adam, Learning rate=1×10−4, Batch size=12026.06 | 11.1641 | 72.95 | 20.2829 | 96 | 1.41 |