Highly Accelerated 3D MRI of Brain Tumors Using Deep Modular Reconstruction Networks
NMR in Biomedicine · 2026-04 · 39 (4)
DOI: 10.1002/nbm.70260 ↗Kurz gefasst
Fachartikel in der Zeitschrift NMR in Biomedicine, online erschienen am 06.03.2026. Verfasst von Anthony Mekhanik und 6 weiteren Autor:innen. Beteiligte Einrichtung: Department of Medical Physics, Memorial Sloan Kettering Cancer Center, New York, New York, USA. Hauptthemen laut MEDLINE: Brain Neoplasms, Imaging, Three-Dimensional, Magnetic Resonance Imaging, Deep Learning. Die Arbeit nennt 29 Literaturangaben. Gefördert von: NCI NIH HHS.
Abstract (Originalsprache)
Post‐contrast 3D T1‐weighted MRI is a time consuming component of cancer neuroimaing protocols. The goal of this study is to accelerate the acquisition of 3D MRI using deep learning reconstruction of undersampled k‐space data beyond clinically‐available acceleration techniques. A two‐stage deep learning framework called Accelerated Deep Modular Image REconstruction (ADMIRE) was developed to first remove aliasing artifacts in the accelerated coronal orientation (unaliasing network) and subsequently enhance magnitude images in axial orientation (enhancement network). The two networks were trained independently using data from two 3D T1‐weighted sequences on GE Healthcare scanners: BRAVO and MPRAGE (N total = 136; N BRAVO = 47; N MPRAGE = 89). 3D MPRAGE k‐space data with eightfold acceleration (4 × 2 k y –k z undersampling) were acquired on 15 patients with enhancing brain lesions and reconstructed with ADMIRE to test feasibility. Validation was performed quantitatively against data‐driven and unrolled deep learning methods and against the clinical acquisition that uses a lower acceleration factor using a qualitative reader study performed by three radiologists. Composite and individual one‐sided Wilcoxon signed‐rank tests were utilized to assess noninferiority of the proposed deep learning approach. Quantitative and qualitative evaluation, scoring and statistical testing revealed that ADMIRE displays better image quality than both data‐driven and unrolled methods and was noninferior compared to the clinical standard, despite 32%–46% reductions in acquisition time, in terms of overall quality (summing all metric scores and using a composite noninferiority margin of 8, p ‐value < 0.001) and diagnostic confidence specifically (using an individual noninferiority margin of 2, p ‐value = 0.004). ADMIRE further increases the current clinical acceleration of T1‐weighted 3D MRI by enabling higher k‐space undersampling factors. The use of a clinical sequence and the fast computation speed facilitate clinical translation of ADMIRE.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Department of Medical Physics, Memorial Sloan Kettering Cancer Center, New York, New York, USA.
- Department of Radiology, Memorial Sloan Kettering Cancer Center, New York, New York, USA.
- Department of Epidemiology & Biostatistics, Memorial Sloan Kettering Cancer Center, New York, New York, USA.
ORCID-Profile: Anthony Mekhanik, Muhammad A Naveed, Ricardo Otazo
Förderung
- NCI NIH HHS
Zitieren
APA
Mekhanik, A., Stember, J. N., Yildirim, O., Naveed, M. A., Zhang, Z., Young, R. J., & Otazo, R. (2026). Highly Accelerated 3D MRI of Brain Tumors Using Deep Modular Reconstruction Networks. NMR in Biomedicine, 39(4). https://doi.org/10.1002/nbm.70260
Vancouver
Mekhanik A, Stember JN, Yildirim O, Naveed MA, Zhang Z, Young RJ, et al. Highly Accelerated 3D MRI of Brain Tumors Using Deep Modular Reconstruction Networks. NMR in Biomedicine. 2026;39(4). doi:10.1002/nbm.70260
BibTeX
@article{Mekhanik2026,
author = {Mekhanik, Anthony and Stember, Joseph N and Yildirim, Onur and Naveed, Muhammad A and Zhang, Zhigang and Young, Robert J and Otazo, Ricardo},
title = {Highly Accelerated 3D MRI of Brain Tumors Using Deep Modular Reconstruction Networks},
journal = {NMR in Biomedicine},
year = {2026},
volume = {39},
number = {4},
doi = {10.1002/nbm.70260}
}
EUVIMED