FONDUE: Robust resolution-invariant denoising of MR images using Nested UNets
Walter Adame-Gonzalez, Aliza Brzezinski-Rittner, Yashar Zeighami, M. Mallar Chakravarty, Reza Farivar, Mahsa Dadar · 2024
Read FONDUE: Robust resolution-invariant denoising of MR images using Nested UNets by Walter Adame-Gonzalez, Aliza Brzezinski-Rittner, Yashar Zeighami, M. Mallar Chakravarty, Reza Farivar, Mahsa Dadar (2024) 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/pmc-12315758.txt.
Description
Walter Adame-Gonzalez and co-authors build a brain scan denoiser that does not care what scanner or resolution produced the image. Their 2024 paper introduces FONDUE, a nested network ensemble trained to clean structural MRI. The training spans three vendors, field strengths from low to ultra-high, and voxel sizes an order of magnitude apart, in healthy and diseased brains alike.
Pushing resolution raises noise, and noise corrupts the measurements morphometry studies live on. FONDUE matched or beat specialist denoisers on every test set on at least one quality metric. It also ran orders of magnitude faster on modest graphics hardware. The code, scripts, and trained weights are public. Large cohort studies process tens of thousands of scans; a fast, resolution-blind cleaner removes a quiet source of bias from all of them.
Catalog record
| CREATOR | Walter Adame-Gonzalez, Aliza Brzezinski-Rittner, Yashar Zeighami, M. Mallar Chakravarty, Reza Farivar, Mahsa Dadar |
|---|---|
| YEAR | 2024 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 87 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |