Two-stage deep learning framework for the restoration of incomplete-ring PET images
2025
Read Two-stage deep learning framework for the restoration of incomplete-ring PET images (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-00816.txt.
Description
Yeqi Fang and Rong Zhou restore images from PET scanners with incomplete detector rings. Missing detectors, whether from failure, cost cuts, or design, can remove about half the coincidence data, and standard reconstruction degrades badly. Their two-stage deep learning framework needs no time-of-flight information.
First, an attention-based network predicts the missing sinogram sections using neighboring slices. After standard reconstruction, a cascaded network and a warm-started diffusion model refine the image. Tests used 613 simulated brain volumes spanning healthy, Alzheimer's, and mild cognitive impairment cases drawn from real scans. Image quality stayed high, with structural similarity above 0.99, roughly double the data loss handled by earlier convolutional methods.
Catalog record
| YEAR | 2025 |
|---|---|
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 31 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-00816