Reproduce “DuSS-PET: Self-Supervised Low-Dose PET Reconstruction Using Pseudo-Dual-Domain Consistency Without Paired Data” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Shuijin Lin, Yuzhu He, Yongbo Wang, Jianhua Ma, Fan Wang
DuSS-PET reconstructs low-dose PET brain slices without paired full-dose scans. It learns from low-dose images and synthetically corrupted versions, aligning image predictions with image-derived pseudo-sinograms. Without fine-tuning on held-out Siemens data, it beats RED at DRF 100: PSNR 25.00 versus 23.89, SSIM 0.8310 versus 0.7139, and NMSE 0.0331 versus 0.0415. Tests use retrospective 2D slices, not scanner-physics validation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.