Super-resolution sodium MRI of human gliomas at 3T using physics-based generative artificial intelligence - Summary - MDSpire

Super-resolution sodium MRI of human gliomas at 3T using physics-based generative artificial intelligence

  • By

  • Catalina Raymond

  • Jingwen Yao

  • Alfredo L. Lopez Kolkovsky

  • Thorsten Feiweier

  • Bryan Clifford

  • Heiko Meyer

  • Xiaodong Zhong

  • Fei Han

  • Nicholas S. Cho

  • Francesco Sanvito

  • Sonoko Oshima

  • Noriko Salamon

  • Linda M. Liau

  • Kunal S. Patel

  • Richard G. Everson

  • Timothy F. Cloughesy

  • Benjamin M. Ellingson

  • June 3, 2025

  • 0 min

Share

Objective:

Enhance sodium MRI quality at 3T for brain tumor imaging using a physics-informed generative adversarial network (GAN) model, with a focus on clinical implications.

Key Findings:
  • Sodium MRI provides unique insights into tumor physiology, particularly in gliomas, with specific metrics demonstrating enhanced resolution.
  • Malignant gliomas show elevated sodium levels linked to increased cellularity and metabolic activity, quantified through imaging analysis.
  • The ATHENA model successfully enhanced the resolution and SNR of sodium MRI images, maintaining biological fidelity as evidenced by validation metrics.
Interpretation:

The use of the ATHENA model demonstrates the potential for improved sodium MRI imaging, which could enhance clinical assessment of brain tumors, particularly in treatment planning and monitoring.

Limitations:
  • Sodium MRI at 3T has low sensitivity and fast T2 decay, limiting spatial resolution; future work should explore methods to mitigate these issues.
  • The model was not trained on native sodium data, which may affect generalizability; further validation on diverse datasets is recommended.
Conclusion:

The ATHENA model represents a promising approach to enhance sodium MRI, potentially improving the evaluation of gliomas and their treatment responses, with implications for clinical practice.

Original Source(s)

Super-resolution sodium MRI of human gliomas at 3T using physics-based generative artificial intelligence

  • by Catalina Raymond, Jingwen Yao, Alfredo L. Lopez Kolkovsky, Thorsten Feiweier, Bryan Clifford, Heiko Meyer, Xiaodong Zhong, Fei Han, Nicholas S. Cho, Francesco Sanvito, Sonoko Oshima, Noriko Salamon, Linda M. Liau, Kunal S. Patel, Richard G. Everson, Timothy F. Cloughesy, Benjamin M. Ellingson

  • June 3, 2025

Related Content

Should checkpoint inhibitors be reserved for biomarker-selected pediatric brain tumors?

Advanced Brain Tumor Care, Close to Home

  • by

  • May 28, 2026

  • 2 min
Glioblastoma survival in rural America: a 10-year experience from a quaternary care center