- Updated: June 12, 2026
- 1 min read
Gradient Step Plug‑and‑Play Model for Dental Cone‑Beam CT Reconstruction
Gradient Step Plug‑and‑Play Model for Dental Cone‑Beam CT Reconstruction
In the rapidly evolving field of dental imaging, reducing photon noise while preserving image quality is a critical challenge. In our latest research, we present a novel gradient‑step plug‑and‑play reconstruction workflow that leverages a data‑driven prior to achieve superior results.
Key Contributions
- Formulation of the reconstruction problem as an inverse problem with a learned prior.
- Simulation of fan‑beam acquisitions with realistic photon noise.
- Training of a gradient‑step denoiser on reconstructed simulated data.
- Integration of the denoiser into a plug‑and‑play algorithm for high‑quality CT reconstruction.
Results
Extensive experiments on synthetic datasets demonstrate the denoising capabilities of the trained model, while qualitative evaluations on real dental cone‑beam CT scans confirm its robustness and generalisation.
Read More
For the full paper and additional resources, visit the arXiv page. Explore related articles and tools on our site: Ubos Tech Blog.
Stay tuned for upcoming updates and implementation guides.
Andrii Bidochko
CTO UBOS
Andrii Bidochko is an AI entrepreneur and researcher focused on AI agents, reinforcement learning, and autonomous systems. He writes about the technologies shaping the future of machine intelligence, from frontier models and agent architectures to real-world AI applications.