Simultaneous Optimization of MP2RAGE T1 -weighted (UNI) and FLuid And White matter Suppression (FLAWS) brain images at 7T using Extended Phase Graph (EPG) Simulations

Abstract

Purpose: The MP2RAGE sequence is typically optimized for either T₁ -weighted uniform image (UNI) or gray matter-dominant fluid and white matter suppression (FLAWS) contrast images. Here, the purpose was to optimize an MP2RAGE protocol at 7 Tesla to provide UNI and FLAWS images simultaneously in a clinically applicable acquisition time at <0.7 mm isotropic resolution.

Methods: Using the extended phase graph formalism, the signal evolution of the MP2RAGE sequence was simulated incorporating T₂ relaxation, diffusion, RF spoiling, and B₁ ⁺ variability. Flip angles and TI were optimized at different TRs (TR_MP2RAGE ) to produce an optimal contrast-to-noise ratio for UNI and FLAWS images. Simulation results were validated by comparison to MP2RAGE brain scans of 5 healthy subjects, and a final protocol at TR_MP2RAGE = 4000 ms was applied in 19 subjects aged 8-62 years with and without epilepsy.

Results: FLAWS contrast images could be obtained while maintaining >85% of the optimal UNI contrast-to-noise ratio. Using TI₁ /TI₂ /TR_MP2RAGE of 650/2280/4000 ms, 6/8 partial Fourier in the inner phase-encoding direction, and GRAPPA factor = 4 in the other, images with 0.65 mm isotropic resolution were produced in <7.5 min. The contrast-to-noise ratio was around 20% smaller at TR_MP2RAGE = 4000 ms compared to that at TR_MP2RAGE = 5000 ms; however, the 20% shorter duration makes TR_MP2RAGE = 4000 ms a good candidate for clinical applications example, pediatrics.

Conclusion: FLAWS and UNI images could be obtained in a single scan with 0.65 mm isotropic resolution, providing a set of high-contrast images and full brain coverage in a clinically applicable scan time. Images with excellent anatomical detail were demonstrated over a wide age range using the optimized parameter set.

Keywords: 7T; FLAWS; MP2RAGE; MRI; ultrahigh field.

FIGURE 1

CNR_total (taking into account the CNRs of UNI and FLAWS_min) obtained from EPG simulations using different TI₁ and TI₂ values for 3 choices of TR_MP2RAGE. Each CNR_total value represented by a colored disk is the maximum found over the range of FAs tested, where for each FA the CNR is the average using all B₁ ⁺ scaling values.  CNR, contrast‐to‐noise ratio; EPG, extended phase graph; FA, flip angle; FLAWS, fluid and white matter suppression; FLAWS_{m i n}, FLAWS minimum image; UNI, uniform image.

FIGURE 2

The INV1 and INV2 mean signal intensities at different TR_MP2RAGEs for different tissues with the uncertainty of the mean, indicated as the error bars using the 3D segmented data from 4 healthy adults

FIGURE 3

The UNI and FLAWS_min CNR plots using different FA combinations at a fixed TI₁/TI₂ of 650/2220 ms and TR_MP2RAGE of 4000 ms. The plots represent the average of simulation results using 10 different B₁ ⁺ values (50% B₁ ⁺ to 140% B₁ ⁺ with steps of 10%). The plots considering individual B₁ ⁺ values are given in Supporting Information Video S1. The upper row shows the CNRs for the UNI image between WM–GM and GM–CSF, and the total CNR. The lower row shows the CNRs for the FLAWS_min image between the GM–WM and GM–CSF, and the total CNR. The crosses correspond to different FA combinations (red: 5°/2°, blue: 5°/4°, green: 5°/5°, orange: 5°/7°). CNR, contrast‐to‐noise ratio; CS, compressed sensing; GM, gray matter; WM, white matter

FIGURE 4

Various contrasts obtained from subject 5 at TR_MP2RAGE = 4000 ms using different α ₂ values

FIGURE 5

The UNI and FLAWS‐related images and the bias fields calculated from the UNI images using (A) the final protocol (B) the MP2RAGE protocol with low sensitivity to B₁ ⁺ by Marques et al. (C) the protocol optimized for FLAWS contrast by Beaumont et al. The details of the protocols are given in Table 2. It is important to note that the final protocol has a shorter TR_MP2RAGE and scan duration and a higher nominal resolution compared to those of the other 2 protocols

FIGURE 6

Three areas (each 40 × 40 mm²) having different B₁ ⁺ values selected from the UNI images that were acquired using (A) the final protocol (B) the MP2RAGE protocol with low sensitivity to B₁ ⁺ by Marques et al. (C) The protocol optimized for FLAWS contrast by Beaumont et al. (Table 2) to demonstrate the achieved contrasts

FIGURE 7

Coronal images with different contrasts acquired at TR_MP2RAGE = 4000 ms from (A) 10‐year‐old healthy child, (B) 62‐year‐old healthy adult, and (C) 12‐year‐old pediatric epilepsy patient with focal cortical dysplasia. The lesion is visible in the left hemisphere (e.g., as a hypo‐intense region indicated by the red arrow in the WM of the UNI image). The nominal image resolution was 0.65 mm isotropic. The scan duration (min:s) was 7:18 for pediatric subjects and 8:58 for the adult subject.