EUVIMEDDescobrir EUVIMED
Open access · CC BY 4.0

Multi‐parametric MRI synthesis for glioblastoma from quantitative MR fingerprinting: Quantitative synthetic neural network (QS‐Net)

Yimin Ni, Chenyang Liu, Wen Li, Ling Lin, Rushan Ouyang, Peng Cao, Ho-Fun Victor Lee, Aya El Helali, Yat-Lam Wong, Xiang Wang, Peilin Wang, Ge Ren and 2 more

Medical Physics · 2026-04 · 53 (4)

DOI: 10.1002/mp.70435 ↗

At a glance

Journal article in Medical Physics, published online on 2026-04-12. Written by Yimin Ni and 13 co-authors. Institution: Department of Health Technology and Informatics, The Hong Kong Polytechnic University, Hung Hom, Hong Kong SAR, China. Major MEDLINE topics: Glioblastoma, Image Processing, Computer-Assisted, Brain Neoplasms, Magnetic Resonance Imaging. 37 references. The full text is freely available (open access, CC BY 4.0). Funded by: National Natural Science Foundation of China Young Scientist, Innovation and Technology Support Program, General Research Fund.

ResumoSubjectsDados bibliográficosCiteRelated studies

Abstract (original language)

Background

Diagnosis and treatment of glioblastoma (GBM) rely on multiparametric MRI (mpMRI), but mpMRI is time‐consuming and costly. Deep learning–based synthesis methods have been proposed to streamline acquisition; however, their generalizability is limited by variability in qualitative input contrasts across sites and scanners.

Purpose

To overcome this limitation, we developed and evaluated a generalizable deep learning model that synthesizes mpMRI contrasts directly from quantitative magnetic resonance fingerprinting (MRF) maps in GBM patients. The proposed Quantitative Synthesis Network (QS‐Net) employs a deeply supervised residual U‐Net generator within an adversarial framework, combined with a two‐stage training strategy to separate anatomical and pathological learning.

Methods

We collected MRF‐derived T1 and T2 maps, along with conventional mpMRI sequences (T1w, T2w, T1‐FLAIR, T2‐FLAIR, and SWI), from 32 healthy volunteers and retrospectively from 18 GBM patient scans. The proposed QS‐Net was initially trained on healthy volunteer data (20 scans for training, 12 for testing) to learn general anatomical features. Subsequently, it was fine‐tuned using 9 GBM patient scans to adapt to pathological characteristics, with the remaining 9 patient scans reserved for independent testing. We compared the performance of QS‐Net against three state of the art deep learning models: Res‐Unet, conditional GAN, and Swin‐Transformer, using both quantitative metrics (MAE, SSIM, and PSNR) and qualitative assessments. Additionally, we assessed the generalizability of the models by evaluating their external validation performance when trained with either conventional MRI or quantitative MRF inputs.

Results

QS‐Net outperformed the comparison models in synthesizing T1w, T2w, SWI, and T2‐FLAIR images for GBM patients, achieving the best results across all quantitative metrics: MAE (1.18 ± 0.52, 1.01 ± 0.36, 1.05 ± 0.37, 1.45 ± 0.76), SSIM (0.934 ± 0.037, 0.939 ± 0.039, 0.934 ± 0.034, 0.926 ± 0.053), and PSNR (29.69 ± 3.21, 29.35 ± 2.29, 29.64 ± 2.58, 27.56 ± 3.39), respectively. Qualitative analysis demonstrated that QS‐Net generated synthetic images with superior resemblance to ground truth, accurately delineating tumor boundaries and preserving intra‐tumoral texture. Furthermore, the generalizability test revealed that models trained on standardized quantitative MRF input maps consistently outperformed models trained on vendor‐specific qualitative MRI inputs across all architectures and metrics ( p < 0.005).

Conclusions

We developed QS‐Net, a deep learning model for high fidelity mpMRI synthesis from quantitative MRF maps, and demonstrated that this quantitative‐input paradigm enables superior cross‐vendor generalization over conventional qualitative MRI‐based approaches.

Subjects

MeSH terms (MEDLINE)

GlioblastomaImage Processing, Computer-AssistedBrain NeoplasmsMagnetic Resonance ImagingNeural Networks, ComputerDeep LearningHumansGenerative Adversarial Networks

Author keywords

contrast synthesismagnetic resonance fingerprintingmodel generalizability

Authors and institutions

  • Department of Health Technology and Informatics, The Hong Kong Polytechnic University, Hung Hom, Hong Kong SAR, China.
  • Department of Radiology, The Eighth Affiliated Hospital, Sun Yat-sen University, Shenzhen, China.
  • Department of Diagnostic Radiology, The University of Hong Kong, Pok Fu Lam, Hong Kong SAR, China.
  • Department of Clinical Oncology, The University of Hong Kong, Pok Fu Lam, Hong Kong SAR, China.
  • Department of Clinical Oncology, Queen Mary Hospital, Pok Fu Lam, Hong Kong SAR, China.

Funding

  • National Natural Science Foundation of China Young Scientist
  • Innovation and Technology Support Program
  • General Research Fund
  • Health and Medical Research Fund

Cite

APA

Ni, Y., Liu, C., Li, W., Lin, L., Ouyang, R., Cao, P., Lee, H. V., Helali, A. E., Wong, Y. L., Wang, X., Wang, P., Ren, G., Cai, J., & Li, T. (2026). Multi‐parametric MRI synthesis for glioblastoma from quantitative MR fingerprinting: Quantitative synthetic neural network (QS‐Net). Medical Physics, 53(4). https://doi.org/10.1002/mp.70435

Vancouver

Ni Y, Liu C, Li W, Lin L, Ouyang R, Cao P, et al. Multi‐parametric MRI synthesis for glioblastoma from quantitative MR fingerprinting: Quantitative synthetic neural network (QS‐Net). Medical Physics. 2026;53(4). doi:10.1002/mp.70435

BibTeX
@article{Ni2026,
  author = {Ni, Yimin and Liu, Chenyang and Li, Wen and Lin, Ling and Ouyang, Rushan and Cao, Peng and Lee, Ho-Fun Victor and Helali, Aya El and Wong, Yat-Lam and Wang, Xiang and Wang, Peilin and Ren, Ge and Cai, Jing and Li, Tian},
  title = {Multi‐parametric MRI synthesis for glioblastoma from quantitative MR fingerprinting: Quantitative synthetic neural network (QS‐Net)},
  journal = {Medical Physics},
  year = {2026},
  volume = {53},
  number = {4},
  doi = {10.1002/mp.70435}
}
Start for free →