Self-Supervised Noise Adaptive MRI Denoising via Repetition to Repetition (Rep2Rep) Learning
Nikola Janjušević, Jingjia Chen, Luke Ginocchio, Mary Bruno, Yuhui Huang, Yao Wang, Hersh Chandarana, Li Feng, N. Janjušević, J. Chen, L. Ginocchio, M. Bruno, Y. Huang, Y. Wang, H. Chandarana, Li Feng · 2025
Read Self-Supervised Noise Adaptive MRI Denoising via Repetition to Repetition (Rep2Rep) Learning by Nikola Janjušević, Jingjia Chen, Luke Ginocchio, Mary Bruno, Yuhui Huang, Yao Wang, Hersh Chandarana, Li Feng, N. Janjušević, J. Chen, L. Ginocchio, M. Bruno, Y. Huang, Y. Wang, H. Chandarana, Li Feng (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-2504-17698.txt.
Description
Nikola Janjušević and co-authors present a self taught method for cleaning noise out of low field MRI scans, in this 2025 paper. Weak scanners are cheaper and more portable, but their images are grainy. Their trick uses the repeated scans that clinics already take: one repeat trains the network to predict another, so no clean reference image is ever needed.
The method also adapts to changing noise levels, a weak point of earlier self taught schemes. On synthetic brain images it matched fully supervised cleaning and beat a leading rival method. On prostate scans from a portable strength scanner, radiologists preferred its cleaned two pass images over raw eight pass images. The authors say the approach fits routine low field practice, where repeats are already on file.
Catalog record
| CREATOR | Nikola Janjušević, Jingjia Chen, Luke Ginocchio, Mary Bruno, Yuhui Huang, Yao Wang, Hersh Chandarana, Li Feng, N. Janjušević, J. Chen, L. Ginocchio, M. Bruno, Y. Huang, Y. Wang, H. Chandarana, Li Feng |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 66 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-17698