Application of Modified Spin-Echo-based Sequences for Hepatic MR Elastography: Evaluation, Comparison with the Conventional Gradient-Echo Sequence, and Preliminary Clinical Experience

Abstract

Purpose To (a) evaluate modified spin-echo (SE) magnetic resonance (MR) elastographic sequences for acquiring MR images with improved signal-to-noise ratio (SNR) in patients in whom the standard gradient-echo (GRE) MR elastographic sequence yields low hepatic signal intensity and (b) compare the stiffness values obtained with these sequences with those obtained with the conventional GRE sequence. Materials and Methods This HIPAA-compliant retrospective study was approved by the institutional review board; the requirement to obtain informed consent was waived. Data obtained with modified SE and SE echo-planar imaging (EPI) MR elastographic pulse sequences with short echo times were compared with those obtained with the conventional GRE MR elastographic sequence in two patient cohorts, one that exhibited adequate liver signal intensity and one that exhibited low liver signal intensity. Shear stiffness values obtained with the three sequences in 130 patients with successful GRE-based examinations were retrospectively tested for statistical equivalence by using a 5% margin. In 47 patients in whom GRE examinations were considered to have failed because of low SNR, the SNR and confidence level with the SE-based sequences were compared with those with the GRE sequence. Results The results of this study helped confirm the equivalence of SE MR elastography and SE-EPI MR elastography to GRE MR elastography (P = .0212 and P = .0001, respectively). The SE and SE-EPI MR elastographic sequences provided substantially improved SNR and stiffness inversion confidence level in 47 patients in whom GRE MR elastography had failed. Conclusion Modified SE-based MR elastographic sequences provide higher SNR MR elastographic data and reliable stiffness measurements; thus, they enable quantification of stiffness in patients in whom the conventional GRE MR elastographic sequence failed owing to low signal intensity. The equivalence of the three sequences indicates that the current diagnostic thresholds are applicable to SE MR elastographic sequences for assessing liver fibrosis. ^© RSNA, 2016.

Figure 1:

MR elastographic images in patients in whom conventional GRE MR elastography was successful (top row) or failed (owing to liver iron overload) (bottom row). Liver is indicated with dotted line. In a typical MR elastography case (top row), magnitude signal intensity is high, shear waves are well visualized on phase and displacement images, and shear stiffness can be successfully calculated within liver on derived stiffness image. In images from the patient with low signal intensity in liver (bottom row), visualization of waves is not possible on phase and displacement images, which results in invalid shear stiffness results. Checkerboard pattern superimposed on displacement and stiffness images represents a mask applied by processing algorithm to areas where phase SNR is below a value that permits reliable evaluation of stiffness.

Figure 2:

Three pulse sequences investigated in this study. A, Conventional GRE MR elastographic sequence. B, SE MR elastographic sequence. C, SE-EPI MR elastographic sequence. TE for each of the three sequences and a schematic of the 60-Hz motion are also shown. Radiofrequency (RF) row indicates the radiofrequency application (excitation and inversion), G_x, G_y, G_z are MR imaging spatial-encoding and motion-encoding gradients, and “Motion” refers to mechanical vibrations introduced by using the MR elastography acoustic driver.

Figure 3:

MR elastographic (MRE) images obtained in patient by using GRE MR elastographic (top row), SE MR elastographic (middle row), and SE-EPI MR elastographic (bottom row) techniques. Shear stiffness maps obtained with the three sequences are similar, demonstrating the equivalence of these three techniques. Checkerboard pattern on elastograms indicates regions where phase SNR is low (inversion confidence <0.95). Liver is indicated with dotted line.

Figure 4:

A, Comparison of shear stiffness values in 130 patients with the three MR elastographic (MRE) sequences. Box-and-whisker plot shows median, first, and third quartile range, and 99th percentile. B, Bland-Altman plot of percentage differences of shear stiffness values obtained with SE and SE-EPI MR elastography (MRE) compared with GRE MR elastography. C, Mean percentage difference between stiffness from SE and SE-EPI sequences compared with GRE data. Confidence intervals for mean percentage differences were −4.35%, −1.14%, and −2.52%, 0.65%, respectively, and are shown with blue diamonds. Both intervals were fully within the 5% equivalence range, which indicates that the three sequences are statistically equivalent.

Figure 5:

MR elastographic (MRE) images in patient with low signal intensity at GRE MR elastography. Images were obtained with GRE MR elastography (top row), SE MR elastography (middle row), and SE-EPI MR elastography (bottom row). Liver is indicated with dotted line. For GRE MR elastography, signal intensity level in liver was low and resulted in noise-dominated phase images and a stiffness map fully masked out because of low inversion confidence. Conversely, both SE sequences had significantly improved SNR and confidence level, with waves clearly visualized on phase-contrast images. Mean hepatic stiffness values were 2.3 kPa ± 0.83 for SE MR elastography and 2.4 kPa ± 0.77 for SE-EPI MR elastography; the GRE data were considered unusable.

Figure 6:

Shear stiffness values obtained with SE and SE-EPI sequences in 47 patients with nonanalyzable GRE MR elastographic images. Stiffness values calculated with the two sequences are equivalent (P < .001), which suggests that the current threshold for fibrosis detection can be used for data obtained with the new acquisitions. According to this threshold, 14 of 47 patients had fibrosis.