نهمین کنگره نقشه برداری مغز

نمایش مقاله

Title:

Extended-MRiLab for Diffusion MRI, Non-Cartesian Acquisition, and Polar Reconstruction

Subject:

Advanced Techniques in Signal Acquisition and Image Processing

Presenter Author:
Name:	Fateme Hoseini Rashkani

Author:
Name 1:	Fateme Hoseini Rashkani - Corresponding - Presenter
Affiliation 1:	Department of Biomedical Engineering, Amirkabir University of Technology

Name 2:	Sara Khamseh
Affiliation 2:	Department of Biomedical Engineering, Amirkabir University of Technology

Name 3:	Abbas Nasiraei Moghaddam
Affiliation 3:	Department of Biomedical Engineering, Amirkabir University of Technology

Name 4:
Affiliation 4:

Name 5:
Affiliation 5:

Name 6:
Affiliation 6:

Name 7:
Affiliation 7:

Name 8:
Affiliation 8:

Name 9:
Affiliation 9:

Name 10:
Affiliation 10:

Abstract

Original Research

introduction :

Magnetic Resonance Imaging (MRI) plays a central role in brain mapping by enabling non-invasive investigation of brain anatomy, microstructure, and connectivity. In particular, diffusion-weighted imaging (DWI) and diffusion tensor imaging (DTI) are essential tools for studying white-matter organization and brain networks. Simulation of MRI experiments is highly valuable in this context, as it allows systematic evaluation of acquisition strategies and reconstruction methods without scanner time constraints or safety concerns. MRiLab is a comprehensive GPU-accelerated MRI simulator that provides flexible tools for sequence design, radiofrequency pulse modeling, virtual object construction, and image reconstruction. Despite its broad capabilities, MRiLab has lacked support for diffusion modeling and advanced non-Cartesian acquisition and reconstruction methods, limiting its applicability to brain mapping research. In this work, we extend MRiLab to support diffusion MRI simulation and introduce a diffusion-weighted signal model for arbitrary gradient directions. Specifically, we implement polar Fourier transform (PFT) and spherical Fourier transform (SFT) reconstruction methods and enable a koosh-ball trajectory for non-Cartesian k-space sampling.

Methods:

All developments were implemented within the MRiLab framework using MATLAB. To simulate diffusion effects, a voxel-wise diffusion tensor model was incorporated. For each voxel, the diffusion-weighted signal was computed as S(b,g)=S₀e^(−b gᵀDg), where b is the diffusion-weighting factor, g is the unit diffusion gradient direction, and D is the diffusion tensor. This formulation enables simulation of DWI and DTI experiments with arbitrary gradient schemes. A whole-brain diffusion tensor dataset (107 × 79 × 60 voxels) was used to construct a virtual brain phantom. Tissue segmentation into white matter, gray matter, and cerebrospinal fluid was performed using the SPM toolbox. Appropriate relaxation parameters and diffusion tensors were assigned to each tissue class, resulting in a realistic brain model for diffusion MRI simulation. To enhance non-Cartesian reconstruction, PFT and SFT methods were implemented alongside the existing non-uniform fast Fourier transform (NUFFT). PFT was designed for 2D radial and 3D stack-of-stars acquisition, while SFT provides a reconstruction framework for 3D radial data acquired using koosh-ball trajectories. In addition, a koosh-ball k-space acquisition macro was developed, enabling direct 3D radial sampling.

Results:

The extended MRiLab framework successfully generated diffusion-weighted images for user-defined b-values and gradient directions. Simulated DWIs demonstrated increased signal attenuation with higher b-values, consistent with theoretical diffusion behavior. Direction-dependent contrast variations were observed when diffusion gradients were applied along orthogonal axes, reflecting anisotropic diffusion in white matter structures. PFT and SFT reconstruction methods were successfully integrated into the MRiLab reconstruction pipeline and made compatible with existing simulation and sequence-design modules. These reconstruction options can now be selected alongside the conventional NUFFT-based approach. In addition, koosh-ball trajectory support was implemented, enabling flexible 3D radial k-space sampling with user-defined parameters.

Conclusion:

This work extends MRiLab into a versatile simulation platform for diffusion MRI and koosh-ball non-Cartesian imaging. By integrating diffusion modeling, PFT and SFT reconstruction methods, and koosh-ball k-space acquisition, the proposed framework enables realistic simulation of brain diffusion experiments and supports methodological development in brain mapping. These enhancements make MRiLab a valuable tool for evaluating diffusion MRI protocols and advanced non-Cartesian imaging strategies prior to in vivo implementation.

Keywords:

Diffusion MRI, MRiLab, Non-Cartesian k-space, Koosh-ball Trajectory, Polar Fourier Transform