Rapid Shear Stiffness Estimations Using 2-D Spatial Excitations in Magnetic Resonance Elastography

Traditional MR elastography (MRE) involves the measurement of 2-D or 3-D shear wave displacement data from which the shear stiffness distribution in an object is estimated. These acquisitions often require 2 to 8 minutes for 2-D data, and as much as 30 minutes for 3-D data. When performed in vivo, patient motion and patient comfort can become an issue with these long scan times. The hypothesis of this work is that rapid shear modulus estimations can be performed using 2-D spatial excitations to acquire 1-D approximations of the shear wave displacement field. Introduction MRE is a phase-contrast technique in which dynamic shear wave motion is encoded as phase in the MR images, and the shear stiffness distribution in an object is estimated from this displacement data [l]. Traditional MRE is performed using a single slice, multiple slices, or a volume excitation to measure 2-D or 3-D displacement data. The imaging sequence is typically repeated with different motion-encoding gradient vector amplitudes and directions ("polarizations") so that the background phase can be removed and motion in different directions can be encoded. Table 1 shows the acquisition times for several different MRE acquisitions. As the number of slices, the number of polarizations, and the TR increase, the 2-D and 3-D acquisition times can become prohibitive for in vivo imaging. Rapid shear stiffness estimates can be performed using a standard GRE MRE sequence with the traditional Fourier spatial encoding replaced by a 2-D Gaussian excitation [2,3]. This produces 1-D displacement data along the length of the excitation "beam" each TR that approximate the true 3-D displacement field. The 1-D displacement data will best approximate the 3-0 displacement data when the beam is oriented perpendicularly to a planar wave front. The I-D displacements can be used to estimate the shear stiffness distribution along the length of the beam. Table 1: Acquisition times for several different MRE acquisitions. Methods MRE data were acquired using a standard 2-D GRE MRE sequence [l] (TWTE = 100/50 ms, 30" flip angle) with a 256x256 acquisition matrix, 20-cm FOV, 8 time offsets, and 400-Hz shear wave motion on a 1.5 T GE Signa scanner. The interrogated object was an agar gel phantom containing two 25-mm diameter cylindrical agar inclusions, a stiffer one on the left and a softer one on the right. The I-D MRE data were acquired using a 2-D excitation approximately 14 mm in diameter. Ten consecutive TRs were averaged together to increase the displacement SNR for each beam. The beams were oriented perpendicularly to the inclusions with one profile acquired through each inclusion and one profile acquired between them. The inversion of the 2-D data to produce an estimate of the 2-D shear stiffness distribution in the object was performed using an algebraic inversion of the differential equation modeling the shear wave propagation (AIDE) [4]. The inversion of the 1-D data was performed using the 1-D analog of the local frequency estimation (LFE) algorithm used by Mandnca et al. [5]. Results The acquired I-D and 2-D displacement data are shown in Figure 1. The I-D wave data can be seen to have the same spatial periodicity as the profiles from the 2-D data. Figure 2 compares the shear stiffness inversions of the 1-D and 2-D data. \. .,