ST-SRPerf: Continuous spatiotemporal representation for perfusion MRI super-resolution through neural ODE and...

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Artificial Intelligence in Medicine

Background:

• Perfusion MRI provides vital data on tissue vascularity for diagnosing conditions like tumors and neurodegenerative diseases.
• Current perfusion MRI techniques face limitations in achieving high spatial and temporal resolution, impacting diagnostic accuracy.
• Inaccuracies in perfusion parameter estimation hinder the full potential of MRI in clinical applications.

Purpose of the Study:

• To introduce a novel framework, ST-SRPerf, for enhancing both spatial and temporal resolution in perfusion MRI.
• To address the inherent trade-offs in traditional perfusion MRI methods.
• To improve image quality and the accuracy of perfusion parameter estimation.

Main Methods:

• Developed SpatioTemporal Super-Resolution framework for Perfusion MRI (ST-SRPerf).
• Integrated neural ordinary differential equations for continuous temporal trajectory calculation.
• Utilized implicit neural representations for continuous spatial implicit representation to achieve super-resolution.

Main Results:

• ST-SRPerf significantly improved image quality in brain and breast perfusion MRI datasets.
• The framework demonstrated superior performance in perfusion parameter estimation compared to existing methods.
• Enhancements in both spatial and temporal resolution were observed across various downsampling factors.

Conclusions:

• ST-SRPerf offers a significant advancement in perfusion MRI super-resolution.
• The framework holds promise for improving diagnostic accuracy in clinical settings.
• This technique provides a valuable tool for enhancing resolution in medical imaging applications.