Self-supervised MRI denoising : leveraging Stein's unbiased risk estimator and spatially resolved noise maps
© 2023. The Author(s)..
Thermal noise caused by the imaged object is an intrinsic limitation in magnetic resonance imaging (MRI), resulting in an impaired clinical value of the acquisitions. Recently, deep learning (DL)-based denoising methods achieved promising results by extracting complex feature representations from large data sets. Most approaches are trained in a supervised manner by directly mapping noisy to noise-free ground-truth data and, therefore, require extensive paired data sets, which can be expensive or infeasible to obtain for medical imaging applications. In this work, a DL-based denoising approach is investigated which operates on complex-valued reconstructed magnetic resonance (MR) images without noise-free target data. An extension of Stein's unbiased risk estimator (SURE) and spatially resolved noise maps quantifying the noise level with pixel accuracy were employed during the training process. Competitive denoising performance was achieved compared to supervised training with mean squared error (MSE) despite optimizing the model without noise-free target images. The proposed DL-based method can be applied for MR image enhancement without requiring noise-free target data for training. Integrating the noise maps as an additional input channel further enables the regulation of the desired level of denoising to adjust to the preference of the radiologist.
| Medienart: |
E-Artikel |
|---|
| Erscheinungsjahr: |
2023 |
|---|---|
| Erschienen: |
2023 |
| Enthalten in: |
Zur Gesamtaufnahme - volume:13 |
|---|---|
| Enthalten in: |
Scientific reports - 13(2023), 1 vom: 19. Dez., Seite 22629 |
| Sprache: |
Englisch |
|---|
| Beteiligte Personen: |
Pfaff, Laura [VerfasserIn] |
|---|
| Links: |
|---|
| Themen: |
|---|
| Anmerkungen: |
Date Revised 23.12.2023 published: Electronic Citation Status PubMed-not-MEDLINE |
|---|
| doi: |
10.1038/s41598-023-49023-2 |
|---|
| funding: |
|
|---|---|
| Förderinstitution / Projekttitel: |
|
| PPN (Katalog-ID): |
NLM366052454 |
|---|
| LEADER | 01000naa a22002652 4500 | ||
|---|---|---|---|
| 001 | NLM366052454 | ||
| 003 | DE-627 | ||
| 005 | 20231227140409.0 | ||
| 007 | cr uuu---uuuuu | ||
| 008 | 231227s2023 xx |||||o 00| ||eng c | ||
| 024 | 7 | |a 10.1038/s41598-023-49023-2 |2 doi | |
| 028 | 5 | 2 | |a pubmed24n1236.xml |
| 035 | |a (DE-627)NLM366052454 | ||
| 035 | |a (NLM)38114575 | ||
| 040 | |a DE-627 |b ger |c DE-627 |e rakwb | ||
| 041 | |a eng | ||
| 100 | 1 | |a Pfaff, Laura |e verfasserin |4 aut | |
| 245 | 1 | 0 | |a Self-supervised MRI denoising |b leveraging Stein's unbiased risk estimator and spatially resolved noise maps |
| 264 | 1 | |c 2023 | |
| 336 | |a Text |b txt |2 rdacontent | ||
| 337 | |a ƒaComputermedien |b c |2 rdamedia | ||
| 338 | |a ƒa Online-Ressource |b cr |2 rdacarrier | ||
| 500 | |a Date Revised 23.12.2023 | ||
| 500 | |a published: Electronic | ||
| 500 | |a Citation Status PubMed-not-MEDLINE | ||
| 520 | |a © 2023. The Author(s). | ||
| 520 | |a Thermal noise caused by the imaged object is an intrinsic limitation in magnetic resonance imaging (MRI), resulting in an impaired clinical value of the acquisitions. Recently, deep learning (DL)-based denoising methods achieved promising results by extracting complex feature representations from large data sets. Most approaches are trained in a supervised manner by directly mapping noisy to noise-free ground-truth data and, therefore, require extensive paired data sets, which can be expensive or infeasible to obtain for medical imaging applications. In this work, a DL-based denoising approach is investigated which operates on complex-valued reconstructed magnetic resonance (MR) images without noise-free target data. An extension of Stein's unbiased risk estimator (SURE) and spatially resolved noise maps quantifying the noise level with pixel accuracy were employed during the training process. Competitive denoising performance was achieved compared to supervised training with mean squared error (MSE) despite optimizing the model without noise-free target images. The proposed DL-based method can be applied for MR image enhancement without requiring noise-free target data for training. Integrating the noise maps as an additional input channel further enables the regulation of the desired level of denoising to adjust to the preference of the radiologist | ||
| 650 | 4 | |a Journal Article | |
| 700 | 1 | |a Hossbach, Julian |e verfasserin |4 aut | |
| 700 | 1 | |a Preuhs, Elisabeth |e verfasserin |4 aut | |
| 700 | 1 | |a Wagner, Fabian |e verfasserin |4 aut | |
| 700 | 1 | |a Arroyo Camejo, Silvia |e verfasserin |4 aut | |
| 700 | 1 | |a Kannengiesser, Stephan |e verfasserin |4 aut | |
| 700 | 1 | |a Nickel, Dominik |e verfasserin |4 aut | |
| 700 | 1 | |a Wuerfl, Tobias |e verfasserin |4 aut | |
| 700 | 1 | |a Maier, Andreas |e verfasserin |4 aut | |
| 773 | 0 | 8 | |i Enthalten in |t Scientific reports |d 2011 |g 13(2023), 1 vom: 19. Dez., Seite 22629 |w (DE-627)NLM215703936 |x 2045-2322 |7 nnns |
| 773 | 1 | 8 | |g volume:13 |g year:2023 |g number:1 |g day:19 |g month:12 |g pages:22629 |
| 856 | 4 | 0 | |u http://dx.doi.org/10.1038/s41598-023-49023-2 |3 Volltext |
| 912 | |a GBV_USEFLAG_A | ||
| 912 | |a GBV_NLM | ||
| 951 | |a AR | ||
| 952 | |d 13 |j 2023 |e 1 |b 19 |c 12 |h 22629 | ||