A Versatile MR Elastography Research Tool with a Modified Motion Signal-to-noise Ratio Approach.

Abstract

Purpose: This study aimed to facilitate research progress in MR elastography (MRE) by providing a versatile and convenient application for MRE reconstruction, namely the MRE research tool (MRE-rTool). It can be used for a series of MRE image analyses, including phase unwrapping, arbitrary bandpass and directional filtering, noise assessment of the wave propagation image (motion SNR), and reconstruction of the elastogram in both 2D and 3D MRE acquisitions. To reinforce the versatility of MRE-rTool, the conventional method of motion SNR was modified into a new method that reflects the effects of image filtering.

Methods: MRE tests of the phantom and liver were performed using different estimation algorithms for stiffness value (algebraic inversion of the differential equation [AIDE], local frequency estimation [LFE] in MRE-rTool, and multimodel direct inversion [MMDI] in clinical reconstruction) and acquiring dimensions (2D and 3D acquisitions). This study also tested the accuracy of masking low SNR regions using modified and conventional motion SNR under various mechanical vibration powers.

Results: The stiffness values estimated using AIDE/LFE in MRE-rTool were comparable to that of MMDI (phantom, 3.71 ± 0.74, 3.60 ± 0.32, and 3.60 ± 0.54 kPa in AIDE, LFE, and MMDI; liver, 2.26 ± 0.31, 2.74 ± 0.16, and 2.21 ± 0.26 kPa in AIDE, LFE, and MMDI). The stiffness value in 3D acquisition was independent of the direction of the motion-encoding gradient and was more accurate than that of 2D acquisition. The masking of low SNR regions using the modified motion SNR worked better than that in the conventional motion SNR for each vibration power, especially when using a directional filter.

Conclusion: The performance of MRE-rTool on test data reached the level required in clinical MRE studies. MRE-rTool has the potential to facilitate MRE research, contribute to the future development of MRE, and has been freely released online.