Introduction

High-b diffusion coefficient D_H of ultrahigh b radial DWI is sensitive to the water permeability at the myelin sheath and can be measured with increased diffusion time using diffusion-weighted stimulated-echo EPI (DWSTEPI) using three 90^o RF pulses ¹. To achieve UHb-DWI with b ~ 10,000 s/mm² on clinical MRI system with limited gradient strength (G_max = 40 mT/m), we varied the mixing time TM up to 450 ms, and permuted slice ordering within the long TM to improve the time-efficiency of the measurement ^2,3. The method requires removing T₁ decay effect on DWSTEPI data and it was unreliable at data with low SNR, such as in b > 5,000 s/mm². On modern MRI system with increased gradient performance (G_max = 80 mT/m), DWI of b ~ 10,000 s/mm² can be obtained using DWSTEPI with a fixed TM (~ 100 ms). However, its time-efficiency is greatly reduced, because a half of the transverse magnetization is wasted, while being restored back to the longitudinal space by the second RF pulse. The fraction of the wasted magnetization is more than 50 % due to the imperfect 90^o flipangle in real situation. Therefore, for successful combination of DWSE and DWSTE, the effect of imperfect 90^o must be corrected, as in previous report ^4,5. In this work, we present a method to obtain UHb-DWI for human CSC using simultaneously measured DWSEPI and DWSTEPI with simple RF correction.

Methods

Pulse Sequence: Manufacturer’s 2D singleshot diffusion-weighted EPI sequence was modified to implement 2D singleshot DW-SESTEPI using a pulse sequence development environment (IDEA, Siemens Medical Solution, Erlangen, Germany). Three diffusion gradients are applied with the identical duration and amplitude. Diffusion-weightings for the low-b DWSE and high-b DWSTE are (γG_Dδ)²(Δ_L-δ/3) and (γG_Dδ)²(Δ_H-δ/3), respectively.
Combining DWSE and DWSTE ^4,5: The magnetizations for DWSE and DWSTE are derived as,
$M_{SE}(r; b_L) =M_+^o (r) \sin {\alpha} \sin^2 {\alpha/2} e^{-TE/T_2} e^{-bD_L}$. ... (1)
$M_{STE}(r; b_H) =M_+^o (r) \sin^3 {\alpha} e^{-TE/T_2} e^{-TM/T_1} e^{-bD_H}$. ... (2)
Therefore, low-b DWSE images can be combined to DWSTE data by multiplying a correction map $g(r)$ as in,
$M_{STE}(r; b_L) = M_{SE}(r; b_L} g(r)$. . . . (3)
Here, the correction map $g(r)$ can be estimated using eqns. (1, 2),
$g(r) = s \cos^2{\alpha/2} e^{-TM/T_1} = M_{STE}(r;b_o)/M_{SE}(r;0)$ . . . (4).
Eq. (4) indicates indicates that the DWSTE signal for the same b-value differs from DWSE by imperfect 90^o and additional T₁ decay.
MR Imaging: UHb-DWI experiment was performed at 3T MRI system (Prisma, Siemens, Erlangen, Germany) using 64 channel head/neck array coil. 2D ss-DW-SESTEPI was applied to a fluid phantom, filled with 0.1 mMol MnCl₂/water, with T₁/T₂ = 950/80 ms, 1x1 mm² in-plane resolution and 4 mm slice thickness, TR 4900 ms, TE 76 ms, 8 averaging, 7 pairs of DWSE and DWSTE with b values ((40, 264), (80, 528), (130, 858), (170, 1123), (210, 1387), (250, 1651)) and ((20, 132), (40, 264), (80, 528)) s/mm² along two orthogonal directions. Then, it was applied to obtain a set of UHb-DW images from CSC of a healthy subject, under an approved IRB, using the identical imaging parameters as those for phantom, except for 7 pairs of DWSE and DWSTE with b values ((250, 1620), (500, 3240), (750, 4860), (1000, 6480), (1250, 8100), (1500, 9721)) and ((130, 842), (250, 1620), (500, 3240)) s/mm² along radial (rDWI) and axial (aDWI) to the cord direction, respectively. Data acquisition time for UHb-DWI was about 6 min for twenty-one slices, covering C1 ~ T2 vertebrae. The measured DWIs are processed, including RF/T1 correction and analyzing signal-b curve, at an offline UNIX computer using a home-developed software, written in Python 3.x.

Results and Discussions

DWSEPI and DWSTEPI images were successfully combined. From the combined images set, diffusion coefficient of the water at room temperature was estimated as D = 2.352 x 10^-3 mm²/s, of which the signal-b curve is illustrated in Fig. 2b. The axial-DWI signal (red symbols in Fig. 2d) of the human CSC fits well to a single-exponential function with respect to b value with diffusivity 2.66x10^-3 mm²/s. However, human radial-DWI signal fits to a double exponential function with average fast and slow decay constants (D_f, D_s) = (0.700, 0.050) x 10^-3 mm²/s and relative fractions (54 %, 46 %). This rDWI signal behavior is very similar to those using the variable TM for UHb-DWI ³. Current method utilizes the entire diffusion-prepared magnetization, unlike the conventional DWSTE. Because the method uses a fixed mixing time and does not require T1 correction, signal-b curve is significantly stable, particularly at the high-b region, say b > 5000 s/mm^2, compared with the method presented in the reference ^2,3.

Synopsis

Ultrahigh-b DWI is obtained using DW-stimulated-echo (DW-STE) sequence, with a cost of a half the prepared magnetization. Therefore, the main objective of this work has been to measure low-b DW-spin-echo and DW-STE in a single sequence, and to correct/combine them for UHb-DWI of cervical spinal cord (CSC).

Acknowledgements

This work was supported by NIH R56 R01 NS106097-01A1.