Clinical acceptance testing and scanner comparison of ultrasound shear wave elastography

Abstract Because of the rapidly growing use of ultrasound shear wave elastography (SWE) in clinical practices, there is a significant need for development of clinical physics performance assessment methods for this technology. This study aims to report two clinical medical physicists’ tasks: (a) acceptance testing (AT) of SWE function on ten commercial ultrasound systems for clinical liver application and (b) comparison of SWE measurements of targets across vendors for clinical musculoskeletal application. For AT, ten GE LOGIQ E9 XDclear 2.0 scanners with ten C1‐6‐D and ten 9L‐D transducers were studied using two commercial homogenous phantoms. Five measurements were acquired at two depths for each scanner/transducer pair by two operators. Additional tests were performed to access effects of different coupling media, phantom locations and operators. System deviations were less than 5% of group mean or three times standard deviation; therefore, all systems passed AT. A test protocol was provided based on results that no statistically significant difference was observed between using ultrasound gel and salt water for coupling, among different phantom locations, and that interoperator and intraoperator coefficient of variation was less than 3%. For SWE target measurements, two systems were compared — a Supersonic Aixplorer scanner with a SL10‐2 and a SL15‐4 transducer, and an abovementioned GE scanner with 9L‐D transducer. Two stepped cylinders with diameters of 4.05–10.40 mm were measured both longitudinally and transaxially. Target shear wave speed quantification was performed using an in‐house MATLAB program. Using the target shear wave speed deduced from phantom specs as a reference, SL15‐4 performed the best at the measured depth. However, it was challenging to reliably measure a 4.05 mm target for either system. The reported test methods and results could provide important information when dealing with SWE‐related tasks in the clinical environment.


| INTRODUCTION
Ultrasound elasticity imaging is one of the most noteworthy technologies developed in ultrasound imaging in the last two decades.
Extensive research efforts have been devoted to the development of various methods, such as compression elastography, transient shear wave imaging, acoustic radiation force imaging, crawling wave imaging, etc. [1][2][3] These methods may also be classified according to the excitation approaches, i.e., the quasi-static or dynamic methods.
Despite the variations, the general idea is to perturb tissue with external or internal mechanical sources to generate a measurable displacement, detect the axial or shear deformation, and then deduce a parameter that is related to tissue elasticity. Either qualitative or quantitative assessment can be achieved. A method is deemed to be quantitative if Young's modulus or shear modulus, or shear wave speed can be directly determined.
Some of these methods, imaging or nonimaging, point or 2D measurements, qualitative or quantitative, have been realized on commercial ultrasound scanners from several vendors. 4,5 Among them, shear wave elastography (SWE) recently has become available on multiple systems and is attractive due to its 2D imaging capability and quantitative nature. Because of the unique tissue mechanical information provided, numerous clinical studies have demonstrated the clinical utility of ultrasound elastography, such as in liver, breast, prostate, kidney, pancreas, and musculoskeletal (MSK) applications. [6][7][8][9][10][11] For example, 2D SWE was shown to have high sensitivity in the assessment of liver fibrosis. 12 The utility of SWE was also demonstrated in assessing the elasticity of normal and pathologic Achilles tendon. 13,14 Because of ultrasound's accessible and affordable nature, many clinical practices are looking into the possibility of providing ultrasound SWE service. Clinical medical physicists are naturally involved in the process of bringing ultrasound elastography into clinical practice. For example, we have faced tasks to perform an acceptance testing (AT) of multiple scanner systems for their SWE function for liver applications, as well as to evaluate different vendors for potential MSK applications of target measurements such as nerves and soft tissue lesions. Despite the intense research developments and clinical applications of SWE functionality itself, few publications exist on the clinical physics aspects of assessing ultrasound elastography performance. 15 In addition, there is no regulatory requirement on acceptance testing or quality assurance on SWE currently. The aim of this work was to report our acceptance testing results and the recommended protocol, as well as our evaluation methods and results of target SWE measurements using systems across vendors.

2.
A | Acceptance testing using homogeneous phantoms Ten new GE LOGIQ E9 XDclear 2.0 scanners (GE Healthcare, Milwaukee, WI, USA), each equipped with a C1-6-D and a 9L-D transducer, were purchased by our practice and included in this study. All scanners and transducers had undergone acceptance testing for physical and mechanical integrity, uniformity, geometric accuracy, depth of penetration, contrast response, and spatial resolution.
All systems passed our acceptance testing. The SWE function on these systems was realized by implementation of the comb-push technology with directional filtering and time-interleaved interpolation of the shear wave tracking, to overcome the lack of software beamforming and low tracking pulse repetition frequency on conventional ultrasound scanners. 16 For AT, we decided to test the performance consistency among the ten systems instead of absolute accuracy, because there appears to be a lack of gold standard in phantom material measurement. 17 Two of the Model 039 shear wave liver fibrosis phantoms were chosen as the test objects (CIRS Inc., Norfolk, VA, USA) ("soft phantom" and "hard phantom" with reference Young's moduli of 3 and 45 kPa, respectively). The reference shear wave speed of the targets can be calculated to be 0.985 and 3.816 m/s using the following equation, assuming a phantom density of 1.03 g/cm 3 as shown in the phantom specification using the fol- where E is the Young's modulus, q is the material density, and c s is the shear wave speed.
Two experienced operators measured the shear wave speed using each of the transducers on its scanner with the two phantoms.
Two imaging depths were obtained (1 and 4 cm for 9L-D, 3 and 7 cm for C1-6-D), because of the known depth dependency of shear wave speed measurements. 18 A circular region of interest (ROI) with an area of 1.8 cm 2 was centered at the abovementioned depths.
The average shear wave speed from the circular ROI was recorded.
Five repeats were obtained for each measurement. Salt water with a concentration of 4.5 g sodium chloride (Sigma-Aldrich, St. Louis, MO, USA) per 100 mL degassed water was used for coupling, as suggested by the RSNA Quantitative Imaging Biomarker Alliance (QIBA) shear wave speed biomarker committee. 19 Transducers were placed in contact with the phantom surface without additional pressure. Acquisition parameters were kept the same among all systems, i.e., push output 100%, track output 100%, shear wave vibration frequency 150-400 Hz for C1-6-D, and 100-500 Hz for 9L-D. All measurements were performed using phantom mode.
Mean and standard deviation (SD) of shear wave speed measurements among all systems are reported. System deviation was assessed as a percentage of the group mean, (max operator (|group mingroup mean|, |group maxgroup mean|))/group mean 9 100.
It was also assessed as multiples of the group SD, (max operator (|group mingroup mean|, |group maxgroup mean|))/SD. System deviation from the group mean value that was less than either 5% or 3SD was deemed to be acceptable in our AT. Related-samples Wilcoxon signed rank test was used to compare the corresponding system shear wave speeds at different depths.
In addition, we also studied several aspects of our AT methodology. In order to assess effects of different coupling media, operators,  The center of these targets was at a depth of 3 cm.

Comparison of target SWE measurements across vendors
Both transaxial and longitudinal measurements were acquired with five repeats for each target on each scanner/transducer pair [ Fig. 1 Among all GE systems with C1-6-D transducers, shear wave speeds were 0.97 AE 0.01 and 1.00 AE 0.01 m/s at 3 cm and 7 cm in the soft phantom, respectively. There was a statistical significant difference between them (P < 0.01). Shear wave speeds were 3.74 AE 0.03 and 3.83 AE 0.10 m/s at 3 and 7 cm in the hard phantom, respectively. There was also a statistical significant difference between them (P < 0.05). A statistically significant difference in shear wave speed measurements was also observed at the two depths in the hard phantom for 9L-D transducers (P < 0.01).
In addition, Fig. 2 depicts the shear wave speed measurements made by using different coupling mediums (ultrasound gel or salt water), three operators, and five fixed probe locations on the T A B L E 1 Mean and standard deviation (SD) of the shear wave speed measurements (m/s) in the acceptance test of ten GE LOGIQ E9 XDclear 2.0 systems. Shallower and deeper depths were 1 and 4 cm, and 3 and 7 cm, for the 9L-D and C1-6-D transducers, respectively.

| DISCUSSION
In this study, we conducted an AT of ten GE LOGIQ E9 XDclear 2.0 systems for liver applications and compared target shear wave measurements of two systems from different vendors. These tasks could be faced by many clinical imaging physicists nowadays. Systemic differences were shown among different transducers and vendors, which were similarly observed in the literature. 18,21,22 For the acceptance tests, all system deviations from group mean were within 5% or 3 SD. Therefore, all systems passed the AT and were accepted for clinical use. Moreover, no statistically significant difference was found between using ultrasound gel or salt water for coupling. Therefore, the AT protocol was updated to not limit to salt water for coupling. Ultrasound gel is used for clinical patients and is more readily available compared to the abovementioned salt water.
Therefore, the number of operators or exact test timing does not This ratio became smaller with smaller target diameters, when using longitudinal plane, or for lower frequency transducers for all scanner/transducer pairs. It should be noted that penetration of the SL15-4 transducer was limited as one would expect (Fig. 4). In addition, band-like artifact regions in the axial direction were also noted on the Aixplorer scanner, 23