Table 2.
Deep learning-based image reconstruction methods available on clinical PET systems
| Manufacturer | Type of reconstruction | Model | PET Tracers | Supported PET System(s) | |
|---|---|---|---|---|---|
| SubtlePET | Subtle Medical | Post-processing noise reduction | Residual + 2.5D U-Net | 18F-FDG, 18F-amyloid, 18F-DOPA, 18F-Choline, 18F-DCFPyL, 68 Ga-DOTATATE, 68 Ga-PSMA | Vendor-neutral |
| AiCE | Canon Medical Systems | Post-processing noise reduction | Residual + DCNN | 18F-FDG | Cartesion Prime |
| HYPER DLR | United Imaging Healthcare | Post-processing noise reduction | Residual + Dense + 2.5D U-Net | 18F-FDG | uMI 550, uMI 780, uEXPLORER |
| HYPER DPR | United Imaging Healthcare | Hybrid | Feedback Net + 2.5D CNN | 18F-FDG | uMI 550, uMI 780, uEXPLORER, uMI Panorama |
| Precision DL | GE HealthCare | Post-processing for non-TOF PET image | Residual + 3D U-Net | 18F-FDG | Omni Legend |