Synopsis

Magnetic resonance acoustic radiation force imaging (MR-ARFI) has been used pre-clinically to detect tissue displacement changes during magnetic resonance guided focused ultrasound treatments. Although displacement is a quantitative metric, shear stiffness is typically used to clinically describe disease states. This work develops and evaluates a model-based technique to determine tissue shear stiffness from MR-ARFI displacement measurements. The MR-ARFI derived values are compared to conventional MR elastography shear stiffness values in a phantom study.

Introduction

Tissue shear stiffness is an inherent tissue property that can be used in both diagnostic and therapeutic medical applications. MR elastography (MRE) is a non-invasive technique that uses phase-difference imaging techniques to determine tissue stiffness through detection of propagating harmonic shear waves generated by an external driver (1). MR acoustic radiation force imaging (MR-ARFI) is another non-invasive technique that can measure internal tissue displacements over a localized region caused by the radiation force of a focused ultrasound beam. While still pre-clinical, MR-ARFI has been used to assess the change of tissue displacement due to thermal therapy (2,3). Although displacement is a quantitative metric that is a surrogate for tissue stiffness, the shear stiffness of the tissue is used clinically to describe disease states (4,5). We have developed a technique to determine tissue shear stiffness using MR-ARFI displacement measurements. This work compares this new technique to MRE measurements using a controlled, tissue-mimicking phantom study.

Methods

Phantoms: Two tissue-mimicking phantoms (cylindrical, 10x15 cm) were fabricated (6) to evaluate the MR-ARFI technique. Two porcine gelatins of different stiffness were used: 125-bloom (Vyse Gelatin Co., IL, USA) and 175-bloom (Sigma-Aldrich Co. MO, USA). One phantom was homogeneous (125-bloom) and the other, a hybrid, had two semicircular compartments composed of the 125- and 175-bloom gelatins (Figure 1). Acoustic properties were measured with through transmission and radiation force balance techniques (6).

MRE was performed using a prototype 2D segmented-EPI based MRE pulse sequence and a pneumatic MRE driver system (Resoundant, Rochester, MN, USA). The mechanical excitation was applied at the base of the phantom and MRE data was acquired coronally to maximize in-plane shear wave propagation. Elastograms were acquired in both phantoms at steady state mechanical excitation frequencies of 30, 40, 50, 60,100 and 150Hz.

MR-ARFI was performed in the phantoms using a pre-clinical MRgFUS system (256-element transducer, 940 kHz, IGT, Inc., Pessac, France). Individual sonications were performed in both phantoms in the 125- (N=6) and 175-bloom (N=3) regions and 3D displacement maps were measured using a simultaneous MR-ARFI and temperature sequence (7). MRE and MR-ARFI setups and relevant parameters are shown in Figure 1 and Table 1.

Shear stiffness calculation: MR images were used to create acoustic models of each phantom. The shear stiffness was determined using an optimization routine that iterated on the shear stiffness to minimize the cost function $$$J = \sum\mid D_{EXP} - D_{SIM} \mid^2$$$ where the sum is over the 3D ROI surrounding the MRgFUS focal spot. D_EXP is the MR-ARFI measured displacement and D_SIM is the simulated MR-ARFI displacement (8). The algorithm was implemented in MATLAB (Mathworks, Natick, MA) and required approximately 2 minutes to converge on a standard PC.

Results

Given the shear wavelengths and the relatively small geometry of the phantoms, MRE derived shear stiffness was estimated by averaging measured shear stiffness in identical ROIs at 100Hz and 150Hz. While the utilization of higher mechanical excitation frequencies were required due to phantom geometry, separate MRE measurements (not shown) of larger phantoms created from the same batches of gelatin demonstrated only minor frequency dependence in stiffness measures over the 30 – 150 Hz mechanical frequency range.
Examples of the wave displacement and elastogram for both the homogenous and dual compartment phantom are shown in Figure 2. An example of the measured and simulated displacement maps for an optimized shear stiffness value in the 125-bloom homogeneous phantom is shown in Figure 3.
As shown in Figure 4, MRE derived shear stiffness values were were 4.88±0.26, 5.03 ±0.19 kPa and 8.73±0.81 kPa in the homogenous 125-bloom, hybrid 125- and 175-bloom compartments respectively versus 5.47±0.64, 6.10±0.90 and 8.50±1.5 kPa for the MR-ARFI derived values.

Discussion & Conclusion

Acute estimation of stiffness change is a potentially valuable parameter when assessing thermal therapies. While MRE is used clinically for estimation of tissue stiffness, it requires a separate mechanical driver that is not easily inserted into the MRgFUS environment. MR-ARFI uses existing MRgFUS hardware, but only measures tissue displacement. We have demonstrated a model-based algorithm that derives accurate shear stiffness from MR-ARFI displacement measurements. These MR-ARFI derived measurements compared well to those derived directly from MRE.
Compared to clinical MRE, MR-ARFI measurements are highly localized. This is not an inherent limitation in the MRgFUS setting as ablation and stiffness assessment pulses are roughly equivalent in size. Further, multi-point MR-ARFI techniques (9) can be applied to interrogate larger areas throughout the treated area. Further studies are necessary to fully characterize this technique and optimize its application in the MRgFUS therapy setting.

Acknowledgements

No acknowledgement found.