Fast water/fatT₂and PDFF mapping via multiple overlapping-echo detachment acquisition and deep learning...

Area of Science:

• Biomedical Engineering
• Medical Imaging
• Neurology

Background:

• Quantitative magnetic resonance imaging (MRI) is crucial for assessing muscle pathology in neuromuscular diseases (NMDs).
• Conventional MRI methods for simultaneous water-fat separation and T2 quantification are time-consuming.
• There is a need for ultrafast MRI techniques to improve diagnostic efficiency for NMDs.

Purpose of the Study:

• To develop and validate an ultrafast MRI method for simultaneous water-fat separation and T2 quantification.
• To assess the diagnostic potential of the novel method for NMDs.

Main Methods:

• A novel chemical shift encoding with the multiple overlapping-echo detachment sequence (CSE-MOLED) framework was developed.
• Deep learning-based reconstruction was employed to address challenges in water-fat separation.
• Experiments were conducted using phantoms and in vivo scans on a 3T MRI scanner, including healthy volunteers and patients with muscle atrophy or damage.

Main Results:

• CSE-MOLED achieved high accuracy (R² > 0.995) and low errors in numerical and phantom experiments for water T2, fat T2, and proton density fat fraction (PDFF).
• The method demonstrated high repeatability (coefficient of variation <2.0%) in phantom and in vivo scans.
• CSE-MOLED successfully differentiated between fat infiltration and muscle damage in patient scans.

Conclusions:

• The ultrafast CSE-MOLED sequence enables simultaneous T2 and proton density mapping for water and fat in 162 ms per slice.
• This technique offers potential for enhanced diagnostic accuracy in NMDs without increasing scan time.
• CSE-MOLED represents a significant advancement in quantitative MRI for muscle pathology assessment.