Introduces diffusion-based generative priors in DGP framework for reconstructing CT images from sparse-view sinograms. Combines iterative optimization with neural generative power while preserving explainability. Shows promising results under highly sparse geometries.
Key Points
- 1.Diffusion models address data scarcity artifacts
- 2.Modifies image generation, model, and optimization
- 3.Maintains model-based explainability
Impact Analysis
Reduces radiation exposure in medical imaging by enabling accurate reconstructions from fewer X-rays. Promising for clinical applications with limited data. Needs further research for broader adoption.
Technical Details
Uses diffusion models in iterative algorithm solving minimization from sinograms. Proposes enhancements to existing DGP approaches. Tested on sparse-angle geometries.