Three-dimensional high-resolution T1 and T2 mapping of whole macaque brain at 9.4 T using magnetic resonance fingerprinting

Abstract

Purpose: Quantitative T₁ and T₂ mapping in non-human primates with whole-brain coverage is challenged by the requirement of sub-millimeter resolution and the inhomogeneity of the transmit magnetic field (B₁⁺ ) covering a large field of view. The goal of the current study is to develop a magnetic resonance fingerprinting (MRF) method for simultaneous T₁ and T₂ mapping of the entire macaque brain within feasible scan time.

Methods: A three-dimensional (3D) MRF sequence with both inversion- and T₂ -preparation modules was developed and evaluated on a 9.4 T preclinical scanner. Data acquisition used a 3D stack-of-spirals trajectory, with undersampling along both the in-plane and the through-plane directions. The effect of B₁⁺ inhomogeneity was accounted for by matching the acquired fingerprint to a dictionary simulated with the B₁⁺ factors measured from a separate scan. In vitro and ex vivo studies were performed to evaluate the accuracy and the undersampling capacity of the MRF method. The application of the MRF method for in vivo, brain-wide T₁ and T₂ mapping was demonstrated on macaques at 4, 6, and 12 years of age.

Results: The MRF method enabled highly repeatable T₁ and T₂ mapping at high spatial resolution (0.35 × 0.35 × 1 mm³ ) with an acceleration factor of 24. In vivo studies showed significant age-related T₂ reduction in deep gray nuclei including the globus pallidus, the putamen, and the caudate nucleus.

Conclusions: This study demonstrates the first MRF study for brain-wide, multi-parametric quantification in non-human primates with sub-millimeter resolution.

Keywords: 9.4 T; macaque brain; magnetic resonance fingerprinting; quantitative MRI; relaxometry.

Figure 1.

The schematics of the MRF sequence (A), the flip angle pattern (B), and the data sampling strategy with 12-fold in-plane and 2-fold through-plane undersampling (C). The colored boxes indicate acquired k_z lines, whereas the dotted boxes indicate k_z lines that were skipped.

Figure 2.

In vitro validation. A. T₁ maps. B. T₂ maps. The red arrow indicates T₂ overestimation. C. ${\text{B}}_{\text{1}}^{\text{+}}$ maps. D&E. Pixel-wise T₁ (D) and T₂ (E) measurements by the conformal coil demonstrating the influence of ${\text{B}}_{\text{1}}^{\text{+}}$ inhomogeneity. Mean T₁ and T₂ values in each compartment measured using the birdcage coil are also shown. The labels of each compartment are indicated in (C). IR-SE: inversion-recovery spin-echo; SE: spin-echo; MRF w/o ${\text{B}}_{\text{1}}^{\text{+}}$: MRF without ${\text{B}}_{\text{1}}^{\text{+}}$ correction; MRF w/ ${\text{B}}_{\text{1}}^{\text{+}}$: MRF with ${\text{B}}_{\text{1}}^{\text{+}}$ correction.

Figure 3.

T₁ and T₂ mapping from undersampled MRF data. A&B. T₁ (A) and T₂ (B) maps from MRF data with varied undersampling schemes. The two numbers indicate the in-plane (${\text{R}}_{\text{xy}}$) and through-plane (${\text{R}}_{\text{z}}$) undersampling factors, respectively. C&D. Normalized root mean square errors of T₁ (C) and T₂ (D) maps with varied undersampling schemes.

Figure 4.

Efficacy of ${\text{B}}_{\text{1}}^{\text{+}}$ correction on ex vivo brain. A&B. Representative T₁ (A) and T₂ (B) maps without (left) and with (right) ${\text{B}}_{\text{1}}^{\text{+}}$ correction. C. ${\text{B}}_{\text{1}}^{\text{+}}$ Map. D. Segmented cerebral and cerebellar cortex. E&F. Distribution of T₁ and T₂ from the cortical ROI in (D) demonstrating the effects of ${\text{B}}_{\text{1}}^{\text{+}}$ correction. MRF w/o ${\text{B}}_{\text{1}}^{\text{+}}$: MRF without ${\text{B}}_{\text{1}}^{\text{+}}$ correction; MRF w/ ${\text{B}}_{\text{1}}^{\text{+}}$: MRF with ${\text{B}}_{\text{1}}^{\text{+}}$ correction.

Figure 5.

Ex vivo validation. A&B. Representative T₁ (A) and T₂ (B) maps in axial (top), sagittal (middle), and coronal (bottom) views. C&D. Comparison of MRF measurements with the conventional methods. Insets show enlarged regions of cerebellum. E. Segmentation of the regions of interest (ROIs). F. Comparison of T₁ (top) and T₂ (bottom) measurements by MRF and the conventional methods in selected ROIs. The ROIs include cerebral cortex (#1, dark blue), caudate nucleus (#2, green), putamen (#3, red), globus pallidus (#4, cyan), thalamus (#5, magenta), geniculate nucleus (#6, light green), hippocampus (#7, yellow), white matter (#8, gray), cerebellar cortex (#9, blue), cerebellar white matter (#10, purple).

Figure 6.

Repeatability of in vivo measurements. A. T₁ maps in three orthogonal views from two repeated MRF scans. B. Pixel-wise comparison of T₁ measurements from repeated scans. C. T₂ maps in three orthogonal views from two repeated MRF scans. D. Pixel-wise comparison of T₂ measurements from repeated scans. The red lines are the lines of identity. R: correlation coefficient.

Figure 7.

Comparison of T₁ and T₂ maps among three age groups. A&B. T₁ (A) and T₂ (B) maps in axial (top), sagittal (middle), and coronal (bottom) views from 4-, 6-, and 12-year-old macaques. C. Comparison of T₂ values in the globus pallidus, the caudate nucleus, and the putamen from each subject, with the regions of interest indicated by the yellow, green, and red arrows in (A&B), respectively.