PURPOSE: Three-dimensional (3D) T₁-weighted sequences such as MP-RAGE are invaluable for evaluation of neonatal and infant brain injury/development. Quantitative assessment of regional tissue volume from 3D MRI images has shown promise in investigating the impact of premature birth. Sequence optimization for neonates has been historically challenging because neonatal brains exhibit reversed white matter–gray matter (WM-GM) contrast on T₁-weighted scans, and the contrast is much lower than that of adult brains. Inspired by Mugler’s optimization strategy [1] for 3D fast spin echo (FSE), we attempted to improve the quality of images acquired with the MP-RAGE sequence by shortening the acquisition train length (ATL; the number of read-out RF rephrasing pulse) using partial-Fourier acquisition. We investigated the effects of ATL on image quality using simulation and in vivo experiments.

METHOD: Simulation: We simulated the effects of ATL on WM-GM contrast efficiency using Bloch’s equation, based on the T₁, T₂, and proton density of the WM, GM and CSF of neonatal brains at 3.0 T: 2840/2170/3700 ms, 266/138/2000 ms, and 0.94/0.90/1.0, respectively [2, 3]. In vivo experiment: Four preterm infants were scanned during natural sleep [2] at 40 weeks corrected age on a 3T Siemens Skyra scanner equipped with a 32-channel head coil using the MP-RAGE sequence with FOV 174 x 192 mm, matrix 174 x 192, number of slices 120, slice thickness 1 mm, TR/TE = 2130/2.98 ms, echo spacing time 8.5ms, slice partial Fourier factors at off, 7/8 and 6/8 settings (corresponding to ATL of 120, 105 and 90, respectively), and flip angle (θ) = 12^o. The total scan time was 3 m and 32s. Evaluation: We estimated SNR = μ_signal/δ_noise, where μ_signal is the mean global tissue intensity (excluding non-brain tissues) and δ_noise is the standard deviation of image noise (assuming noise is Gaussian distributed) [5]. Similarly, CNR = (μ_GM-μ_WM)/δ_noise where μ_GM and μ_WM are the mean tissue intensities of the GM and WM, respectively. The performance of the optimization method was then evaluated using SNR and CNR efficiencies, defined as SNR and CNR per square root of total scan time (TA) (second), SNReff = SNR /(TA^0.5) and CNReff = CNR /(TA^0.5).

RESULT AND DISCUSSION: Figure 1 shows the theoretical effective inversion recovery time for the three ATLs with the optimized protocol proposed in [2, 4]. The results indicated that the corresponding for ATL of 120, 105 and 90 were around 1300, 1500 and 1600 ms, respectively. The WM-GM contrast efficiency increased by approximately 40% when the ATL decreased from 120 to 90. In vivo scans (Figure 2) showed that both signal intensity and contrast increased with decreasing ATL. Further quantitative analysis found that SNR efficiency (Figure 3a) increased from 5.9 to 11.1 (by 88%) and CNR efficiency (Figure 3b) increased from 1.18 to 2.11 (by 79%) when ATL decreased from 120 to 90. Student’s paired t-test showed that the improvements were highly significant (P = 0.003 and P<0.001, respectively). The simulation showed that contrast efficiency increased with decreasing ATL, but underestimated the magnitude of the increase. Potential reasons for the underestimate may include: (1) The MR parameters of the neonatal brains (T₁ & T₂ relaxation times and proton density) exhibits variability between term and preterm babies, across different brain regions of a given subject and across subjects [6]. Because of the lack of more precise data, we were unable to account for this variability and used the average neonatal MR parameters in the simulation. (2) Without loss of generality, we simplified the simulation by ignoring k-space configuration. However, in in vivo experiments, we conducted k-space optimization by shifting k-space zero filling [2, 4] to obtain the optimal image quality.

CONCLUSION: In this study, we show in preterm neonates that shortening the acquisition train length of the MP-RAGE sequence significantly improved SNR and CNR efficiencies. The proposed optimization methodology can be easily extended to other populations (e.g. term infants, adults and elders), and different organs, field strengths and MR sequences.