Improving temporal resolution in fMRI using a 3D spiral acquisition and low rank plus sparse (L+S) reconstruction

Area of Science:

• Medical Imaging
• Neuroscience
• Biophysics

Background:

• Rapid whole-brain dynamic Magnetic Resonance Imaging (MRI) is crucial for Blood Oxygen Level Dependent (BOLD) functional MRI (fMRI).
• Faster fMRI acquisitions enhance sensitivity, allow physiological noise filtering, and reduce motion artifacts.
• Undersampling in faster MRI leads to aliasing artifacts, necessitating advanced reconstruction techniques.

Purpose of the Study:

• To evaluate the feasibility of the low-rank (L) plus sparse (S) matrix decomposition (L+S) method for accelerating block-design fMRI.
• To assess the L+S method's ability to separate background signals from BOLD signals in accelerated fMRI data.
• To demonstrate the L+S method's effectiveness with a 3D stack of spirals (SoS) acquisition technique.

Main Methods:

• Applied the L+S decomposition model to reconstruct dynamic MRI data from undersampled k-space.
• Utilized a 3D stack of spirals (SoS) acquisition with kz-t domain undersampling for block-design fMRI.
• Acquired and analyzed retrospective and prospective fMRI data from human volunteers at 3T during a visual task.

Main Results:

• The L+S method successfully separated temporally correlated background information (L component) from sparse BOLD signals (S component).
• Achieved whole-brain coverage of 40 slices at 2mm isotropic resolution with a 64x64 matrix size every 500ms.
• Demonstrated the potential of SoS fMRI with L+S reconstruction at an acceleration factor of four.

Conclusions:

• The L+S method is a viable technique for accelerating fMRI acquisitions, particularly for block-design paradigms.
• This acceleration strategy improves temporal sampling, crucial for capturing dynamic BOLD responses.
• The findings support the use of L+S reconstruction for high-temporal-resolution, whole-brain fMRI.