Magnetic Resonance Elastography of kidneys: SE-EPI MRE reproducibility and its comparison to GRE MRE

Abstract

The purpose of this study is 1) to demonstrate reproducibility of spin echo-echo planar imaging (SE-EPI) magnetic resonance elastography (MRE) to estimate kidney stiffness; and 2) to compare SE-EPI MRE and gradient recalled echo (GRE) MRE-derived stiffness estimations in various anatomical regions of the kidney. Kidney MRE was performed on 33 healthy subjects (8 for SE-EPI MRE reproducibility and 25 for comparison with GRE MRE; age range: 22-66 years) in a 3 T MRI scanner. To demonstrate SE-EPI MRE reproducibility, subjects were scanned for the first scan and then asked to leave the scan room and repositioned again for the second (repeat) scan. Similar set-up was used for GRE MRE as well. The displacement data was then processed to obtain overall stiffness estimates of the kidney. Concordance correlation analyses were performed to determine SE-EPI MRE reproducibility and agreement between GRE MRE and SE-EPI MRE derived stiffness. A high concordance correlation (ρ_c = 0.95; p-value<0.0001) was obtained for SE-EPI MRE reproducibility. Good concordance correlation was observed (ρ_c = 0.84; p < 0.0001 for both kidneys, ρ_c = 0.91; p < 0.0001 for right kidney and ρ_c = 0.78; p < 0.0001 for left kidney) between GRE MRE and SE-EPI MRE derived stiffness measurements. Paired t-test results showed that stiffness value of medulla was significantly (p < 0.0001) greater than cortex using SE-EPI MRE as well as GRE MRE. SE-EPI MRE was reproducible and good agreement was observed in MRE-derived stiffness measurements obtained using SE-EPI and GRE sequences. Therefore, SE-EPI can be used for kidney MRE applications.

Keywords: MR Elastography of kidneys; kidney MRE; kidney stiffness.

Figure 1:

Schematic of experimental setup. A passive driver consisting of two soft pads was placed on subject’s posterior side connected to the active driver placed outside the scanner room to induce 60 Hz vibrations into the kidney.

Figure 2:

Schematic of MRE Sequences used in this study. (a) Flow compensated SE-EPI MRE sequence and (b) Rapid GRE MRE sequence.

Figure 3:

Detailed MRE data processing framework using in-house custom built Matlab code.

Figure 4:

This shows the high resolution T1-weighted images used to draw ROI for (a) renal cortex, (b) renal medulla.

Figure 5:

This shows magnitude image of the kidneys (a, f), snapshot of displacement field in the three spatial encoding directions (i.e. x, y, z), respectively at same time points (b-d, g-i), and the corresponding overall weighted stiffness map (e,j) for both kidneys obtained using SE-EPI MRE and GRE MRE respectively.

Figure 6:

Concordance correlation plot comparing all anatomical regions (cortex and medulla) of both kidneys. The correlation coefficient is 0.95 with a significant p-value<0.0001.

Figure 7:

Bland-Altman’s plot showing good agreement between Scan 1 and Scan 2 using SE-EPI MRE with a mean bias of −0.079 kPa and narrow limits of 95% confidence interval (CI): −0.48 kPa – 0.323 kPa thereby demonstrating reproducibility of SE-EPI MRE sequence.

Figure 8:

Concordance correlation plots showing good correlation in stiffness values obtained using GRE MRE and SE-EPI MRE sequences for all anatomical regions (cortex, and medulla) of (a) both kidneys with correlation coefficient of 0.84 with a significant p-value<0.0001); (b) the right kidney with correlation coefficient of 0.91 with a significant p-value<0.0001; and (c) the left kidney with correlation coefficient of 0.78 with a significant p-value<0.0001.

Figure 9:

Bland-Altman’s plots showing good agreement between stiffness values obtained using GRE MRE and SE-EPI MRE for (a) both kidneys with a mean bias of 0.21 kPa and narrow limits of 95% confidence interval (CI): −0.34 kPa – 0.75 kPa; (b) right kidney with a mean bias of 0.15 kPa and narrow limits of 95% confidence interval (CI): −0.29 kPa – 0.58 kPa; and (c) left kidney with a mean bias of 0.27 kPa and limits of 95% confidence interval (CI): −0.35 kPa – 0.89 kPa.

Figure 10:

Box plots showing difference in stiffness values of cortex and medulla using SE-EPI MRE and GRE MRE (all with a significant p-value<0.0001).

Figure 11:

First harmonic amplitude map showing in one subject for both SE-EPI and GRE MRE data. Significant (p<0.0001) increase in SE-EPI MRE first harmonic amplitude was found when compared to GRE MRE.