Self‐supervised learning for denoising of multidimensional MRI data
Magnetic Resonance in Medicine · 2024-11 · 92 (5) · 1980-1994
DOI: 10.1002/mrm.30197 ↗At a glance
Journal article in Magnetic Resonance in Medicine, published online on 2024-06-27. Written by Beomgu Kang and 4 co-authors. Institution: School of Electrical Engineering, Korea Advanced Institute of Science and Technology, Daejeon, Republic of Korea. Major MEDLINE topics: Magnetic Resonance Imaging, Signal-To-Noise Ratio, Brain, Image Processing, Computer-Assisted. Cited 16 times so far (Crossref); 53 references. The full text is freely available (open access, CC BY-NC-ND 4.0). Funded by: NIBIB NIH HHS, NINDS NIH HHS, National Research Foundation of Korea (NRF) grant funded by Korea government (MSIT).
Abstract (original language)
Purpose
To develop a fast denoising framework for high‐dimensional MRI data based on a self‐supervised learning scheme, which does not require ground truth clean image. Theory and Methods Quantitative MRI faces limitations in SNR, because the variation of signal amplitude in a large set of images is the key mechanism for quantification. In addition, the complex non‐linear signal models make the fitting process vulnerable to noise. To address these issues, we propose a fast deep‐learning framework for denoising, which efficiently exploits the redundancy in multidimensional MRI data. A self‐supervised model was designed to use only noisy images for training, bypassing the challenge of clean data paucity in clinical practice. For validation, we used two different datasets of simulated magnetization transfer contrast MR fingerprinting (MTC‐MRF) dataset and in vivo DWI image dataset to show the generalizability.
Results
The proposed method drastically improved denoising performance in the presence of mild‐to‐severe noise regardless of noise distributions compared to previous methods of the BM3D, tMPPCA, and Patch2self. The improvements were even pronounced in the following quantification results from the denoised images.
Conclusion
The proposed MD‐S2S (Multidimensional‐Self2Self) denoising technique could be further applied to various multi‐dimensional MRI data and improve the quantification accuracy of tissue parameter maps.
Subjects
MeSH terms (MEDLINE)
Author keywords
Authors and institutions
- School of Electrical Engineering, Korea Advanced Institute of Science and Technology, Daejeon, Republic of Korea.
- Bionics Research Center, Biomedical Research Division, Korea Institute of Science and Technology (KIST), Seoul, Republic of Korea.
- Department of Radiology and Biomedical Imaging, Yale School of Medicine, New Haven, Connecticut, USA.
- Divison of MR Research, Department of Radiology, Johns Hopkins University, Baltimore, Maryland, USA.
ORCID profiles: Beomgu Kang, Wonil Lee, Hye-Young Heo, HyunWook Park
Funding
- NIBIB NIH HHS
- NINDS NIH HHS
- National Research Foundation of Korea (NRF) grant funded by Korea government (MSIT)
- NIH HHS
- KIST Institutional Programs
- Korea Medical Device Development Fund grant funded by the Korea government
- National Institutes of Health
Cite
APA
Kang, B., Lee, W., Seo, H., Heo, H. Y., & Park, H. (2024). Self‐supervised learning for denoising of multidimensional MRI data. Magnetic Resonance in Medicine, 92(5), 1980-1994. https://doi.org/10.1002/mrm.30197
Vancouver
Kang B, Lee W, Seo H, Heo HY, Park H. Self‐supervised learning for denoising of multidimensional MRI data. Magnetic Resonance in Medicine. 2024;92(5):1980-1994. doi:10.1002/mrm.30197
BibTeX
@article{Kang2024,
author = {Kang, Beomgu and Lee, Wonil and Seo, Hyunseok and Heo, Hye-Young and Park, HyunWook},
title = {Self‐supervised learning for denoising of multidimensional MRI data},
journal = {Magnetic Resonance in Medicine},
year = {2024},
volume = {92},
number = {5},
pages = {1980--1994},
doi = {10.1002/mrm.30197}
}
EUVIMED