Study of ultrasound stiffness imaging methods using tissue mimicking phantoms

Highlights

• We study the elastographic appearance of different types of solid lesions using agar based tissue mimicking phantoms.

• Characterization of acoustic and elastic properties of the phantoms is done using pulse echo method and compression test.

• We do imaging of phantoms using Ultrasound Elastography and Acoustic Radiation Force Impulse (ARFI) Imaging.

• We report a comparative study of Ultrasound Elastography and ARFI with B-mode imaging based on image metrics.

• We validate the results with clinical Elastography patient images.

Abstract

A pilot study was carried out to investigate the performance of ultrasound stiffness imaging methods namely Ultrasound Elastography Imaging (UEI) and Acoustic Radiation Force Impulse (ARFI) Imaging. Specifically their potential for characterizing different classes of solid mass lesions was analyzed using agar based tissue mimicking phantoms. Composite tissue mimicking phantom was prepared with embedded inclusions of varying stiffness from 50 kPa to 450 kPa to represent different stages of cancer. Acoustic properties such as sound speed, attenuation coefficient and acoustic impedance were characterized by pulse echo ultrasound test at 5 MHz frequency and they are ranged from (1564 ± 88 to 1671 ± 124 m/s), (0.6915 ± 0.123 to 0.8268 ± 0.755 db cm^-1 MHz^-1) and (1.61×10⁶ ± 0.127 to 1.76 × 10⁶ ± 0.045 kg m^-2 s^-1) respectively. The elastic property Young’s Modulus of the prepared samples was measured by conducting quasi static uni axial compression test under a strain rate of 0.5 mm/min upto 10 % strain, and the values are from 50 kPa to 450 kPa for a variation of agar concentration from 1.7% to 6.6% by weight. The composite phantoms were imaged by Siemens Acuson S2000 (Siemens, Erlangen, Germany) machine using linear array transducer 9L4 at 8 MHz frequency; strain and displacement images were collected by UEI and ARFI. Shear wave velocity 4.43 ± 0.35 m/s was also measured for high modulus contrast (18 dB) inclusion and X.XX m/s was found for all other inclusions. The images were pre processed and parameters such as Contrast Transfer Efficiency and lateral image profile were computed and reported. The results indicate that both ARFI and UEI represent the abnormalities better than conventional US B mode imaging whereas UEI enhances the underlying modulus contrast into improved strain contrast. The results are corroborated with literature and also with clinical patient images.

Introduction

Ultrasound B-mode imaging is a popular and most widely used method for imaging breast, thyroid, prostate and human abdominal organs like kidney, spleen and liver. Even though it is used as a screening tool in cancer diagnosis, it is poor at distinguishing cancerous tissue from soft tissue. Basically there are two types of cancer tissue namely benign and malignant depending on whether or not they can spread by invasion and metastasis. Benign lesions are those that cannot spread out by invasion. They grow only locally and they can be cured by suitable therapy where as malignant tumor invades neighboring cells, enter into blood vessels, lymphatic system and metastasize to different sites. For distinguishing benign and malignant lesions, conventional ultrasound techniques use B-mode image shape features like lesion margin irregularity, shadowing, microlobulation and wider than taller orientation. However these features are often found to be overlapping, which decreases the reliability of B-mode in classification of lesions [1]. This leads to invasive biopsies to confirm the presence of cancer which causes patient discomfort and unnecessary anxiety. On the other hand, pathological cancerous changes of tissues are highly correlated with changes in stiffness [2]. Abnormalities such as benign and malignant cancer lesions could be identified based on their stiffness properties; benign tumors are generally around 2–3 times stiffer than normal tissues and deform more for an applied compression. But malignant tumors are harder than surrounding tissues and show less deformation. Thus they can be distinguished by stiffness contrast than acoustic contrast and this property is used in stiffness imaging. In recent decades, there has been an increasing need in assessing the stiffness properties like Young’s modulus E and shear modulus G of tissues. In isotropic materials, the ratio of longitudinal deformation (strain) in response to an applied longitudinal force (stress) is known as Young’s modulus (E) of elasticity. The shear modulus (G) relates transverse strain to transverse stress. A number of stiffness imaging modalities are being developed and they are based on applying a mechanical excitation to tissues of interest and measuring tissue deformation. The measured deformation can be displayed directly as an image or strain is computed and displayed as a grey scale map known as elastogram [3].

Stiffness imaging methods are categorized according to the source of mechanical excitation and how the local displacements are measured. Mechanical excitation may be either ’external’ via probe (static or quasi static) [3], ’organic’ which relies on natural movements of the body like heart beat, pulsing of blood vessels [4], [5], ’dynamic’ external vibration to create shear waves within the tissue of interest called as sonoelasticity [6] or locally by acoustic radiation force [7]. Stiffness imaging methods can be further classified based on the method of deformation measurement either by ultrasound [3], [4], [5], Magnetic Resonance Imaging [8] or optical methods [9]. The focus of this paper is on Ultrasound Elastography Imaging (UEI) which is based on external compression and Acoustic Radiation Force Impulse (ARFI) [10], [11], [12] imaging which is based on force generated by ultrasound.

Initial clinical results were obtained for identifying breast [13], thyroid [14], [15] and liver [16] using UEI and ARFI imaging. However, issues like amount of compressive force to be applied to get repeatability in Ultrasound Elastography, image contrast representations between tumor and background, complex nature of tissue structures and data on elastic properties of normal and pathological tissues need through evaluation [17], [18]. In clinical setting, Ultrasound Elastography and ARFI imaging are still not standardized as B-mode imaging in terms of instrumentation parameters like optimum gain setting, dynamic range, and depth of penetration. In order to do this, there is a need for a large database of stiffness properties of different categories of human tissue. It would be difficult to obtain tissues exhibiting malignant or benign features at will. Moreover biological tissues lose their characteristics with time when they are harvested from body. It is hence necessary to develop tissue mimicking phantoms which have identical acoustic (speed of sound, attenuation coefficient and acoustic impedance) and elastic properties (Young’s modulus and shear modulus) of human soft and cancerous tissues. Tissue mimicking phantoms should be temporally stable; when they are tested diagnostically for validation and training purposes, the geometry, acoustic and elastic properties should be maintained within a tolerable range of 1–2% or it should be at least predictable [19], [20]. The commonly used physical gels for tissue like phantoms are agar and gelatin [19], [21] where as chemical gels are polyacrylamide [22] and polyvinyl alcohol [9]. Since physical gel phantoms are made by hydrogels, there is a chance for bacterial invasion and reduction of water content as time passes through. Proper care must be taken to avoid the desiccation of water from phantoms. Chemical gel phantoms might be more stable than physical gel phantoms [23], however physical gel phantoms are easier, safer and can be made stable for a longer duration (more than 6 months) either by adding preservatives or with suitable preconditioning. [21].

Having said all these, the objective of this work is to study the performance of UEI and ARFI imaging by quantifying how well the underlying elastic modulus contrast is represented for various categories of inclusions made in tissue mimicking phantoms. As yet, no studies have compared the performance of conventional Ultrasound B mode, UEI and ARFI in custom made tissue mimicking phantoms. The performance is evaluated by finding out the parameters namely Contrast Transfer Efficiency (CTE) and lateral image profile which is the pictorial representation of contrast distribution in image. Lateral image profile gives the degree with which the inclusion is differentiated locally from its surrounding tissue whereas Contrast Transfer Efficiency gives how well the actual contrast (elastic contrast of inclusion to background) is depicted in image. In order to achieve this, agar based homogeneous phantoms were made and their acoustical properties were characterized using pulse echo ultrasound test. Mechanical properties were measured by conducting uni axial compression test on Universal Testing Machine (UTM). Thereafter a composite phantom with embedded inclusions were prepared and stiffness parameters were imaged using Siemens Acuson S2000 (Siemens, Erlangen, Germany) machine. Both qualitative and quantitative results were obtained. The resultant strain images were pre-processed and parameters were extracted to assess the performance. The paper is organized as follows. In Section 2, we explain about the development of phantoms and their characterization based on measurement of acoustic and elastic properties. Then we present design, development of composite phantom and the imaging techniques. In Section 3, the results are discussed.