Toward Super-Resolution Reconstruction of Diffusion-Relaxation MRI Using Slice Excitation With Random Overlap (SERO).

Area of Science:

• Medical Imaging
• Biophysics
• Neuroimaging

Background:

• Diffusion MRI is crucial for probing tissue microstructure.
• Low signal-to-noise ratio (SNR) and resolution limit diffusion MRI's effectiveness.
• Accurate estimation of microstructural parameters like diffusivity and relaxation times is challenging.

Purpose of the Study:

• Introduce Slice Excitation with Random Overlap (SERO) for improved diffusion MRI.
• Enable variable repetition times (TRs) and diffusion weighting within a single acquisition shot.
• Support super-resolution reconstruction of key microstructural parameters (S0, D, V, T1).

Main Methods:

• Implemented a diffusion-weighted spin-echo sequence with pseudo-random slice excitation overlap in Pulseq.
• Achieved variable TRs (0.15-21.9s) and high b-values (up to 1.4 ms/μm²).
• Evaluated accuracy and precision in numerical phantoms and demonstrated feasibility in vivo at 1.5-mm isotropic resolution in 2:30 min.

Main Results:

• SERO improved accuracy for diffusivity (D), diffusional variance (V), and T1 estimation in simulations.
• Achieved high precision comparable to direct sampling across various SNRs.
• Regularized SERO significantly reduced RMSE at low SNR (SNR=3) compared to direct sampling (SNR ≥ 7-10).
• In vivo scans showed sharp tissue boundaries and smooth parameter maps.

Conclusions:

• Random slice overlap enriches encoding diversity, enhancing diffusion and relaxation parameter accuracy and precision.
• SERO provides a novel approach for high-resolution microstructural imaging.
• This technique is particularly beneficial for low SNR environments, improving diagnostic capabilities without extending scan duration.