EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Clean Self-Supervised MRI Reconstruction from Noisy, Sub-Sampled Training Data with Robust SSDU

Charles Millard iD, Mark Chiew iD

Bioengineering · 23.12.2024 · 11 (12) · 1305

DOI: 10.3390/bioengineering11121305 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Bioengineering, online erschienen am 23.12.2024. Verfasst von Charles Millard und 1 weiteren Autor:innen. Beteiligte Einrichtung: Wellcome Centre for Integrative Neuroimaging, FMRIB, University of Oxford, Oxford OX3 9DU, UK. Die Arbeit wurde bisher 12-mal zitiert (Crossref) und nennt 53 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Wellcome Trust, Wellcome Trust Core Award, Royal Academy of Engineering.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Most existing methods for magnetic resonance imaging (MRI) reconstruction with deep learning use fully supervised training, which assumes that a fully sampled dataset with a high signal-to-noise ratio (SNR) is available for training. In many circumstances, however, such a dataset is highly impractical or even technically infeasible to acquire. Recently, a number of self-supervised methods for MRI reconstruction have been proposed, which use sub-sampled data only. However, the majority of such methods, such as Self-Supervised Learning via Data Undersampling (SSDU), are susceptible to reconstruction errors arising from noise in the measured data. In response, we propose Robust SSDU, which provably recovers clean images from noisy, sub-sampled training data by simultaneously estimating missing k-space samples and denoising the available samples. Robust SSDU trains the reconstruction network to map from a further noisy and sub-sampled version of the data to the original, singly noisy, and sub-sampled data and applies an additive Noisier2Noise correction term upon inference. We also present a related method, Noiser2Full, that recovers clean images when noisy, fully sampled data are available for training. Both proposed methods are applicable to any network architecture, are straightforward to implement, and have a similar computational cost to standard training. We evaluate our methods on the multi-coil fastMRI brain dataset with novel denoising-specific architecture and find that it performs competitively with a benchmark trained on clean, fully sampled data.

Schlagwörter

Stichwörter der Autor:innen

deep learningimage reconstructionmagnetic resonance imaging

Autor:innen und Institutionen

  • Wellcome Centre for Integrative Neuroimaging, FMRIB, University of Oxford, Oxford OX3 9DU, UK.
  • Department of Medical Biophysics, University of Toronto, Toronto, ON M4N 3M5, Canada.
  • Physical Sciences, Sunnybrook Research Institute, Toronto, ON M4N 3M5, Canada.

ORCID-Profile: Charles Millard, Mark Chiew

Förderung

  • Wellcome Trust
  • Wellcome Trust Core Award
  • Royal Academy of Engineering
  • EPSRC
  • Engineering and Physical Sciences Research Council
  • NIHR Oxford BRC
  • Canada Research Chairs Program

Zitieren

APA

Millard, C., & Chiew, M. (2024). Clean Self-Supervised MRI Reconstruction from Noisy, Sub-Sampled Training Data with Robust SSDU. Bioengineering, 11(12), 1305. https://doi.org/10.3390/bioengineering11121305

Vancouver

Millard C, Chiew M. Clean Self-Supervised MRI Reconstruction from Noisy, Sub-Sampled Training Data with Robust SSDU. Bioengineering. 2024;11(12):1305. doi:10.3390/bioengineering11121305

BibTeX
@article{Millard2024,
  author = {Millard, Charles and Chiew, Mark},
  title = {Clean Self-Supervised MRI Reconstruction from Noisy, Sub-Sampled Training Data with Robust SSDU},
  journal = {Bioengineering},
  year = {2024},
  volume = {11},
  number = {12},
  pages = {1305},
  doi = {10.3390/bioengineering11121305}
}
Kostenlos starten →