EUVIMEDEUVIMED entdecken
Crossref

Toward safe deployment of deep learning in MRI: A physics-driven uncertainty framework for automated quality control and risk Tiering in virtual fat suppression

Rong Liu, Zenghan Zhou, Yingying Song, Zongyan Jiang, Yuanfeng Liu

Magnetic Resonance Imaging · 2026

DOI: 10.1016/j.mri.2026.110700 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Magnetic Resonance Imaging, 2026. Verfasst von Rong Liu und 4 weiteren Autor:innen. Beteiligte Einrichtung: Wuhan Puren Hospital (Affiliated to Wuhan University of Science and Technology), 1 Benxi Street, Jianshe Fourth Road, Qingshan District, 430080 Wuhan…. Hauptthemen laut MEDLINE: Deep Learning, Magnetic Resonance Imaging, Adipose Tissue, Image Processing, Computer-Assisted. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 36 Literaturangaben. Gefördert von: Hubei Key Laboratory of Mine Environmental Pollution Control and Remediation, Natural Science Foundation of Hubei Province.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

PURPOSE: Deep generative models in MRI are hindered by "hallucinations" and a lack of safety mechanisms. This study introduces a physics-driven framework for trustworthy Virtual Fat Suppression (VFS), enabling automated quality control and active risk tiering. METHODS: A differentiable Bloch layer was embedded into an RRDB-RaGAN architecture to enforce physics-based signal consistency during synthesis. An uncertainty-guided training strategy with artifact-enriched supervision was further introduced to support a three-tier risk model for auto-pass, human review, and rejection. The framework was evaluated using quantitative fidelity metrics, real-world severe artifact detection, and blinded clinical reader assessment. RESULTS: In real-world severe artifact detection, the image-level uncertainty score achieved an AUC of 0.9053 and a recall of 81.82% at a false-positive rate of 4.69%. In blinded clinical evaluation, agreement between AI-predicted and reader-mapped risk tiers was substantial (quadratic weighted Cohen's κ = 0.7938), and images assigned to the Low-Risk tier achieved a mean Likert score of 4.70 ± 0.61. CONCLUSION: By coupling physics-constrained synthesis with uncertainty-based risk governance, the proposed framework provides a practical and auditable safety layer for virtual fat suppression in clinical MRI workflows.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Deep LearningMagnetic Resonance ImagingAdipose TissueImage Processing, Computer-AssistedHumansQuality ControlArtifactsUncertaintyAlgorithmsPhysicsAutomation

Stichwörter der Autor:innen

Automated quality controlMagnetic resonance imagingPhysics-informed deep learningTrustworthy AIUncertainty quantification

Autor:innen und Institutionen

  • Wuhan Puren Hospital (Affiliated to Wuhan University of Science and Technology), 1 Benxi Street, Jianshe Fourth Road, Qingshan District, 430080 Wuhan, Hubei, China; Institute of Medical Innovation and Transformation, Pu…
  • Wuhan Puren Hospital (Affiliated to Wuhan University of Science and Technology), 1 Benxi Street, Jianshe Fourth Road, Qingshan District, 430080 Wuhan, Hubei, China.
  • Institute of Medical Innovation and Transformation, Puren Hospital Affiliated to Wuhan University of Science and Technology, 1 Benxi Street, Jianshe Fourth Road, Qingshan District, 430080 Wuhan, Hubei, China; Wuhan Liu…

Förderung

  • Hubei Key Laboratory of Mine Environmental Pollution Control and Remediation
  • Natural Science Foundation of Hubei Province

Zitieren

APA

Liu, R., Zhou, Z., Song, Y., Jiang, Z., & Liu, Y. (2026). Toward safe deployment of deep learning in MRI: A physics-driven uncertainty framework for automated quality control and risk Tiering in virtual fat suppression. Magnetic Resonance Imaging. https://doi.org/10.1016/j.mri.2026.110700

Vancouver

Liu R, Zhou Z, Song Y, Jiang Z, Liu Y. Toward safe deployment of deep learning in MRI: A physics-driven uncertainty framework for automated quality control and risk Tiering in virtual fat suppression. Magnetic Resonance Imaging. 2026. doi:10.1016/j.mri.2026.110700

BibTeX
@article{Liu2026,
  author = {Liu, Rong and Zhou, Zenghan and Song, Yingying and Jiang, Zongyan and Liu, Yuanfeng},
  title = {Toward safe deployment of deep learning in MRI: A physics-driven uncertainty framework for automated quality control and risk Tiering in virtual fat suppression},
  journal = {Magnetic Resonance Imaging},
  year = {2026},
  doi = {10.1016/j.mri.2026.110700}
}
Kostenlos starten →