Simultaneous T2, T2*, and R2' Mapping for Multiple Sclerosis Using Nonlinear Model-Based Reconstruction of Undersampled Radial RARE-EPI MRI.

Abstract

Purpose: To demonstrate the synergy of undersampled radial 2in1-RARE-EPI acquisition and nonlinear model-based reconstruction for accelerated and simultaneous T₂, T₂*, and R₂' mapping in brains of patients with multiple sclerosis (MS).

Methods: 2in1-RARE-EPI combines a RARE module with an EPI module to capture T₂ and T₂* information. Nonlinear model-based reconstruction was applied to estimate T₂, T₂* maps directly from undersampled k-space data. A retrospective undersampling experiment was conducted to compare nonlinear model-based and parallel imaging compressed sensing (PICS) reconstruction. The proposed approach was validated and compared to reference methods multiecho spin-echo (T₂, MSE) and multiecho gradient-echo (T₂*, MGRE) in a phantom, healthy subjects, and MS patients.

Results: 2in1-RARE-EPI together with nonlinear model-based reconstruction enabled T₂, T₂*, and R₂' mapping with 7.5-fold scan-time acceleration relative to the references, while addressing key limitations of reference techniques, including long acquisition times, misregistration, motion and off-resonance sensitivity, and the need for calibration scans. Phantom and in vivo validation showed that the parametric maps obtained with this approach were in agreement with the reference methods. Compared with PICS, nonlinear model-based reconstruction showed more consistent spatial detail and accuracy at higher acceleration factors. The proposed method detected small focal lesions in T₂, T₂*, and R₂' maps of MS patients and enabled visualization of the central vein sign.

Conclusion: Scan time reduction facilitated by nonlinear model-based reconstruction of 2in1-RARE-EPI provides a technical foundation for enhanced patient compliance, and is a fundamental precursor for broader clinical studies on the potential of T₂, T₂*, and R₂' as imaging biomarkers.