Ultrasound elastography for imaging tendons and muscles

Abstract

Ultrasound elastography is a recently developed ultrasound-based method which allows the qualitative or quantitative evaluation of the mechanical properties of tissue. Strain (compression) ultrasound elastography is the commonest technique performed by applying mild compression with the hand-held transducer to create real-time strain distribution maps, which are color-coded and superimposed on the B-mode images. There is increasing evidence that ultrasound elastography can be used in the investigation of muscle, tendon and soft tissue disease in the clinical practice, as a supplementary tool to conventional ultrasound examination. Based on preliminary data, potential clinical applications include early diagnosis, staging, and guiding interventions musculotendinous and neuromuscular disease as well as monitoring disease during rehabilitation. Ultrasound elastography could also be used for research into the biomechanics and pathophysiology of musculotendinous disease. Despite the great interest in the technique, there is still limited evidence in the literature and there are several technical issues which limit the reproducibility of the method, including differences in quantification methods, artefacts, limitations and variation in the application of the technique by different users. This review presents the published evidence on musculoskeletal applications of strain elastography, discusses the technical issues and future perspectives of this method and emphasizes the need for standardization and further research.

Introduction

Ultrasound elastography (EUS) is an ultrasoundbased method for the assessment of the mechanical properties of tissue^(1–3). The method was initially introduced in vitro in the ‘90s and later evolved into an imaging tool for in vivo applications^(1–5). EUS is based upon the principle that stress applied to tissue causes changes within it, which depend on the elastic properties of tissue^(1–5). There are several approaches of EUS, depending on the method to assess elasticity^(1–3). Free hand strain (compression) EUS is the commonest technique that has been employed mainly in the field of oncology^(6–12).

Degenerative and traumatic muscle and tendon disease results in biomechanical changes. Since the commercial availability of EUS, there has been a lot of interest for potential clinical applications of the method, such as for early diagnosis of muscle and tendon disease, for guiding and monitoring therapy, for predicting the risk of injury in athletes and for assessing the effect of physiotherapy or training in healthy and diseased musculotendinous tissue^(13–29).

This review aims at presenting the main indications of EUS for muscle and tendon disease based on the published evidence, at presenting the technical difficulties and limitations of the method and finally at discussing the future perspectives of EUS for assessing the musculoskeletal system.

Free hand strain (compression) EUS

Depending on the way of stress application and displacement detection, there are several EUS techniques, including strain EUS, shear wave EUS, transient EUS and acoustic radiation force EUS^{(2, 3, 14)}. The most commonly used method is free-hand strain EUS, also found in the literature as compression elastography, sonoelastography, and real-time elastography^(17–22). This is performed by manually compressing the tissue using the hand-held US transducer^(17–22).

Strain EUS is based upon Hook's law for the calculation of Young's elastic modulus (E), which is a physical quantity measuring elasticity⁽¹⁾. By assuming that the applied stress is uniform, the elastic moduli are inversely proportional to the measured strain (E = stress/strain)⁽¹⁾. Strain represents the amount of deformation and can be calculated is as the change in distance between two points (displacement) divided by the initial distance⁽¹⁾. So the main principle of strain (compression) EUS is that a compressive force (stress) is applied to tissue causing axial tissue displacement, which is lower in hard tissue and higher in softer tissue^(1–4). The displacement data are used to construct strain distribution maps (elastograms) (fig. 1)^(1–4). The elastogram is superimposed on the conventional B-mode image as a gray scale or color-coded image, displayed next to the conventional B-mode image on the screen in real time. Although the gray/color scale encoding is elective by the user, usually red is used for encoding soft tissue, blue for hard tissue and yellow/green for tissue of intermediate stiffness (fig. 1). The elastogram is a relative image, where the strain of each area is displayed relative to the strain of other tissues within the region of interest. There is also a semiquantitative measurement method (the strain ratio), which represents the ratio of the strain of the area of interest (ROI) to an equally measuring area in the reference tissue (usually fat).

Fig. 1.

Longitudinal (a, c) and transverse (b, d) strain elastograms of the middle third of asymptomatic Achilles tendons (T) showing two distinct EUS patterns: type 1 tendons (c, d) appear homogenously stiff (green/blue); type 2 tendons (a, b) appear considerably inhomogenous with soft (red) areas, which do not correspond to any changes in B-mode US. The retroachilles fat (F) appears as a mosaic of green, red and blue. Note the red areas at the lateral and medial sides of the tendon in the transverse plane (b, d) which correspond to artifacts, secondary to difficulty in stabilizing the transducer

US elastography for the evaluation of tendons

Most of the clinical research on musculoskeletal applications of strain EUS focuses on the Achilles tendon. It has been found that the normal Achilles tendons in healthy volunteers may present with two distinct EUS appearances: They may be either homogenously hard structures or, in the majority of cases (62%), they may be considerably inhomogenous with soft areas parallel to the long axis of the tendon (longitudinal bands or spots) (fig. 1)⁽¹⁷⁾. The areas of distinct softening may not correspond to any changes in B-mode or Doppler US⁽¹⁷⁾. The above findings are confirmed by two studies comparing normal (asymptomatic) and abnormal (symptomatic) tendons^(18–20). The majority of asymptomatic tendons were hard (86–93%), however, in 1.3–12% cases there was mild or discrete softening (orange or red areas)^(18–20). On the contrary, symptomatic tendons were characterised by discrete softening in 57% and mild softening in 11%⁽¹⁸⁾. Mild softening (yellow) did not correlate to B-mode US abnormalities, whereas discrete softening (red) was found mainly in cases with US pathology^{(19, 20)}. Therefore, it has been suggested that only discrete soft areas (red) in Achilles tendon should be considered as abnormal^{(19, 20)} (fig. 2). It is not yet clear what the alterations in asymptomatic and sonographically normal tendons represent; it is suggested that they may either correspond preclinical changes or to false positive findings, possibly secondary to tissue shifting at interfaces between collagen fibers^(17–20). However, a study of 12 athletes with Achilles tendinopathy evaluated using US, EUS and MRI reported completely different findings⁽²¹⁾. This study reports increased stiffness in the abnormal tendons⁽²¹⁾. It seems that there is some discrepancy on the appearance of normal Achilles tendons between the studies available so far, which points out the need for further research on the field.

Fig. 2.

Longitudinal free hand strain elastogram of a 34-year-old recreational runner with healed Achilles tendon injury. There is hypoechogenicity at the superficial half of the Achilles tendon at the area of the healed partial injury. The abnormal area appears softer (yellow with red areas) compared to the stiffer (blue/green) normal-appearing remaining tendon. The Kager's fat appears as a mosaic of various levels of stiffness

Free-hand strain EUS for the Achilles tendon has been found to have high accuracy and good reproducibility^(17–21). The overall correlation between US and EUS findings was found to be excellent (accuracy 97%), if mild softening (yellow) is considered physiological and only distinct red areas are considered abnormal^(18–20). Compared to clinical findings, EUS has a mean sensitivity of 93.7% and a specificity of 99.2%⁽¹⁸⁾. The inter- and intra-observer reproducibility of the EUS was found to be good to excellent, if the evaluation of the Achilles tendon elastogram was performed qualitatively^{(17, 21)} and poor (variation 29–37%), if semiquantitative software measurements (strain ratio) were employed⁽¹⁷⁾.

Besides Achilles tendon disease, lateral epicondylitis has also been evaluated in a single study, which showed significant softening in the abnormal extensor tendons with strong correlation with US and clinical findings⁽²²⁾. Further applications of EUS for tendon disease based on preliminary data and the authors’ personal experience include patella tendinopathy, regenerated tendons after harvesting and rotator cuff tendinopathy and tears; however, there are no casecontrolled studies available in the published literature on these applications yet (figs. 3, 4)^(30–32).

Fig. 3.

Longitudinal free-hand gray-scale (b) and color-coded (c) strain elastogram and corresponding B-mode image (a) of a 22-year-old football player with proximal patella tendinopathy. The tendinopathic area appears hypoechoic (a) and softer than the remaining tendon (b, d). In this case soft is encoded as white (gray-scale) or blue (color scale)

Fig. 4.

B-mode and free-hand strain elastograms of a normal supraspinatus tendon footplate (a) and a case with partial articular surface tear of the supraspinatus tendon (b). The normal tendon appears hard (blue) and can be easily differentiated from the normal bursa which appears softer (red). The tear (*) appears as a softer (yellow/red) area within the harder (blue) remaining tendon (b). Note the presence of soft (red) lines beneath the greater tuberosity, corresponding to artifacts

US elastography for the evaluation of muscles

There are limited data available on the use of strain SEL for normal and diseased skeletal muscle. EUS is a feasible and accurate means to produce muscle elasticity maps and evaluate normal skeletal muscle⁽²³⁾. Normal relaxed muscle is found to be an inhomogeneous mosaic of intermediate stiffness (green and yellow colors) with scattered softer and harder areas especially at the periphery near boundaries (fig. 5)^{(26, 27)}. However, it has not been investigated yet what the color variation depends on. A few studies have showed differences in the elasticity (strain index) of the masseter muscles between sexes as well as differences in elasticity of periocular rectus medialis and lateralis muscles in various gaze positions^{(24, 25)}.

Fig. 5.

Transverse free hand strain elastograms of normal relaxed vastus lateralis muscle, which appears as a mosaic of intermediate or increased stiffness (green or blue color respectively) with scattered softer (red) areas near muscle boundaries. The subcutaneous fat appears soft (red/yellow)

Degenerative and neuromuscular disease has been studied using strain EUS. A study on inflammatory myositis showed changes in elasticity of the affected muscles (increased or reduced stiffness, due to fibrosis or fatty infiltration respectively)⁽²⁶⁾. A correlation between quantitative strain EUS parameters and elevated serum markers was also found and the study concluded that EUS may be helpful in staging and monitoring inflammatory myopathies⁽²⁶⁾. Similar findings are reported in a case study of a patient with congenital Bethlem myopathy, where exact correlation between EUS and US/MR imaging findings was found⁽²⁷⁾. EUS also detected changes which were not evident on US and MR, possibly indicating increased sensitivity for myopathic changes⁽²⁷⁾. A study using vibration and Doppler signal alterations reported that EUS could be used for depicting myofascial trigger points⁽²⁸⁾. In cases of cerebral palsy spasticity, EUS has been found useful to detect changes in contracted muscle and to establish the site of botolinum toxin injection⁽²⁹⁾. Based on the above data, it seems that EUS has a role in the early diagnosis and staging of dystrophic, myopathic and spastic muscle and in guiding intervention in muscle disease.

Although there are preliminary reports on the use of EUS for diagnosis and staging of rotator cuff muscle atrophy and for depicting and staging muscle injury, there is no published evidence yet^(30–32).

Technical issues, limitations and artifacts of EUS

One of the major issues lies in the fact that there are various EUS techniques and processing algorithms available, therefore, the findings as well as the artifacts or limitations may be highly dependent on the technique used and may be specific to a certain system. Most of the experience on technical problems and ways to overcome them has resulted from the use of free-hand compression EUS. Compression EUS is technically very challenging in terms of producing artifact-free cineloops of decompression-compression cycles. A major issue is the correct amount of pressure to be applied, which ideally should be moderate, as in high or low levels of pressure the elastic properties of tissue become non-linear, and thus the calculation of strain is not correct⁽⁶⁾. Most EUS systems now provide an on-screen indicator providing real-time feed back on the appropriate amount of pressure. To minimize intra-observer variation, the scoring or measurements should be based on evaluation of images derived after reviewing entire cineloops^(17–20). Another major problem in strain EUS is the lack of quantitative measurements, leading to the need of alternative methods for the evaluation of the elastograms, including semiquantitative measurements either using the built-in software (strain ratio)⁽¹⁷⁾ or using external computer software^{(21, 25, 26)} or qualitative visual evaluation of elastograms^(17–20). This has led to lack of reproducibility and difficulty in comparing the results among different studies.

When using EUS for examining musculoskeletal tissue, special issues should be taken into consideration. Anisotropy should be avoided, as the B-mode appearance influences the acquisition of EUS data^(18–20). When examining the Achilles tendon, longitudinal images are of better quality than transverse images, because of artifacts at the medial and lateral sides of the image⁽¹⁷⁾. When examining a long structure (e.g. the Achilles tendon), overlapping images should be acquired, to overcome the problem of artifacts at the borders of the elastogram caused by inhomogenous pressure^(17–20). For the evaluation of soft tissue masses, EUS may be challenging in cases of superficial protuberant masses, where the application of uniform compression may be difficult.

The size of the elastogram should be taken into consideration when comparing elastograms. The elastogram is a relative image where the elasticity of each tissue is displayed compared to the mean elasticity of all tissues. This fact is not a major issue in breast, as the surrounding tissue is fairly homogenous (fat and glandular tissue). In musculoskeletal EUS though, the elastogram may include tissues with wider elasticity differences (fat, tendon, bone, muscle), and thus wider scatter in acquired elasticity data. For the Achilles tendon, the suggested, but not universally applied, standard size is a depth of 3 times the tendon and about 3/4 of the screen in longitudinal scans. In transverse scans the paratenon should be included in the elastogram⁽²⁰⁾.

In many musculoskeletal applications (e.g. Achilles tendon), the tissue of interest is very superficial. In most US systems a minimum distance from surface (usually about 1 mm) is needed to place the elastogram, so the use of gel pads or probe adaptors is necessary in such cases to increase the distance between the skin and probe^(18–21). In conventional musculoskeletal imaging, the use of large amounts of gel is common practice in order to create an even surface and to reduce the amount of pressure on the tissue. However, when performing EUS for musculoskeletal applications, care should be taken not to include the gel in the elastogram, as it results in dramatic changes in the elastogram, making the tendons appear considerably stiffer compared to the gel (fig. 6).

Fig. 6.

The impact of including gel in the strain elastograms. Axial strain elastograms of the same asymptomatic Achilles tendon including (a) and excluding (b) small amount of gel. The inclusion of some gel in the elastogram results in a homogenously stiffer tendon without areas of distinct softening (red), which are evident there is no gel included (b). The EUS settings and the level of pressure were kept stable, as indicated on the screen

Several artifacts may lead to misinterpretation of the images. These include fluctuant changes at the edges of the elastogram^(17–19), red (soft) lines around calcifications, behind dense bone, at the superficial margin of homogenous lesions⁽²⁰⁾ and at the interfaces between tissues (e.g. between adjacent muscles). Characteristic artifacts are also associated with cystic masses, which appear as a mosaic of all levels of stiffness and with lesions adjacent to major vessels present, due to pulsations. The above artifacts should be excluded from the qualitative or quantitative scoring of the elastograms.

Future perspectives

EUS probably represents the most important technical development in the field of ultrasonography since Doppler imaging. The advantages of EUS include the low cost, and non-invasiveness and the potential of wider clinical availability than other methods of elasticity such as MR elastography. So far there is promising evidence that EUS can be used to assess the mechanical properties of musculoskeletal tissues in the clinical setting and that EUS may even be more sensitive than MR or US in detecting subclinical changes of muscle and tendon. Therefore EUS could potentially be valuable for early diagnosis, for monitoring during rehabilitation medicine and as a research tool for biomechanics and pathophysiology of musculotendinous disease.

Limitations include the limited amount of published evidence available and several technical issues limiting the reproducibility of the method, including lack of quantification, artifacts and variation in the application of the technique between users. Moreover, considerable criticism focuses on its potential clinical utility, as in most cases EUS showed changes already evident on conventional US or Doppler imaging, whereas EUS changes not evident on conventional imaging were clinically occult.

For all the above reasons, we think that a more systematic approach for the investigation of this new diagnostic tool should be undertaken, including standardization of EUS for soft tissue applications, selecting the proper indications for research and exploring newer software algorithms. Standardization is of paramount importance in order to achieve consistency in the application of the technique, which would allow for comparisons between findings. The indications for EUS should be established and would ideally focus on symptomatic but non-US evident disease or very early stages of disease, in order to investigate whether EUS is more sensitive than conventional imaging. Newer algorithms such as shear wave EUS or ARFI and quantitative EUS should be evaluated in comparison to strain EUS.

Conclusion

EUS in its current form remains a highly subjective technique with debatable clinical value, due to lack of standardization and limited evidence. Further multicenter controlled studies are needed including large populations and long term follow up together with correlation with histology, conventional imaging, biomechanical and clinical data, in order to describe the pattern and clinical significance of EUS findings. With proper standardization and further structured research, EUS may become a valuable supplementary tool in the investigation of musculoskeletal disease.