Quantitative Ultrasound Techniques Used for Peripheral Nerve Assessment

Aim: This review article describes quantitative ultrasound (QUS) techniques and summarizes their strengths and limitations when applied to peripheral nerves. Methods: A systematic review was conducted on publications after 1990 in Google Scholar, Scopus, and PubMed databases. The search terms “peripheral nerve”, “quantitative ultrasound”, and “elastography ultrasound” were used to identify studies related to this investigation. Results: Based on this literature review, QUS investigations performed on peripheral nerves can be categorized into three main groups: (1) B-mode echogenicity measurements, which are affected by a variety of post-processing algorithms applied during image formation and in subsequent B-mode images; (2) ultrasound (US) elastography, which examines tissue stiffness or elasticity through modalities such as strain ultrasonography or shear wave elastography (SWE). With strain ultrasonography, induced tissue strain, caused by internal or external compression stimuli that distort the tissue, is measured by tracking detectable speckles in the B-mode images. In SWE, the propagation speed of shear waves, generated by externally applied mechanical vibrations or internal US “push pulse” stimuli, is measured to estimate tissue elasticity; (3) the characterization of raw backscattered ultrasound radiofrequency (RF) signals, which provide fundamental ultrasonic tissue parameters, such as the acoustic attenuation and backscattered coefficients, that reflect tissue composition and microstructural properties. Conclusions: QUS techniques allow the objective evaluation of peripheral nerves and reduce operator- or system-associated biases that can influence qualitative B-mode imaging. The application of QUS techniques to peripheral nerves, including their strengths and limitations, were described and discussed in this review to enhance clinical translation.


Introduction
Ultrasound (US) is an increasingly popular modality for imaging peripheral nerves in the clinic, providing important information about nerve microstructure [1,2]. Many peripheral nerves are located superficially and, hence, are easily accessible for US examination. Although neurophysiological assessments are often considered the gold standard for peripheral nerve assessment [3], US-based techniques can be used as a complementary tool, providing anatomical information and localizing lesions more specifically [4], and providing additional data to guide clinical interventions [4][5][6][7]. Using US for peripheral nerve evaluation is advantageous compared to other imaging modalities, particularly in pediatric [4,8] and geriatric [9,10] populations, due to its rapid scanning capability and wide in-clinic availability.
The dominant US imaging technique in medical studies is reflection or pulse-echo technology (echo-ultrasonography), which analyzes the signals returned to the transducer US devices and transducers are often categorized based on their nominal operating frequencies into the following three ranges: 1-15 MHz is typical for current clinical scanners; 15-30 MHz is typically designated as high-frequency ultrasound (HFUS); and 30-100 MHz is typically designated as ultra-high-frequency ultrasound (UHFUS) [16][17][18]. HFUS and UHFUS developments have resulted in improved image resolution and quality when evaluating superficial tissues in the human body [13,19,20] such as peripheral nerves [21,22], eye [19,23], and skin [13,17,18,20]. Commonly utilized clinical HFUS transducers for nerve imaging operate with peak frequencies of >20 MHz, which produce very high-resolution US images at shallow depths < 5 cm. In current clinical practice, peripheral nerves are evaluated qualitatively using grey-scale B-mode images, which provide information about nerve structure, and using color Doppler images to show vascularity [1,[5][6][7].
US is a promising modality for assessing the peripheral nerve status, but it is limited by its qualitative basis as well as system-and operator-dependency [12]. Thus, the qualitative interpretation of US images is often variable or inconclusive. For example, neuropathy often leads to altered nerve echogenicity and even the disappearance of fascicular architecture in B-mode images. Increased vascularity, abnormal anatomical structures, and reduced nerve mobility, indicative of tethering, can also be considered in a neuropathy diagnosis with US [5][6][7]. However, such qualitative nerve characterizations may vary for different operators and imaging setups. Morphometric assessments of B-mode images may provide more objective metrics for evaluating nerve structure. For example, the increased cross-sectional area (CSA) of median nerves has often been correlated with the diagnosis of carpal tunnel syndrome (CTS) [5,24]. To account for differences in patient anthropometric characteristics, dimensionless CSA ratios of nerves (e.g., carpal tunnel inlet-to-outlet) have also been used to assess neuropathy [25]. Nevertheless, even these indices may be confounded by operator-induced and platform-specific sources of variability. Moreover, such morphometric measures lack information about the nerves' microstructure. Figure 2 shows B-mode images on the transverse plane (short axis plane) of the median nerve in an exemplary healthy participant compared with a CTS patient. Remarkably, the fascicles detected in the healthy median nerve cannot be seen in the CTS patient [5]. US is a promising modality for assessing the peripheral nerve status, but it is limited by its qualitative basis as well as system-and operator-dependency [12]. Thus, the qualitative interpretation of US images is often variable or inconclusive. For example, neuropathy often leads to altered nerve echogenicity and even the disappearance of fascicular architecture in B-mode images. Increased vascularity, abnormal anatomical structures, and reduced nerve mobility, indicative of tethering, can also be considered in a neuropathy diagnosis with US [5][6][7]. However, such qualitative nerve characterizations may vary for different operators and imaging setups. Morphometric assessments of B-mode images may provide more objective metrics for evaluating nerve structure. For example, the increased crosssectional area (CSA) of median nerves has often been correlated with the diagnosis of carpal tunnel syndrome (CTS) [5,24]. To account for differences in patient anthropometric characteristics, dimensionless CSA ratios of nerves (e.g., carpal tunnel inlet-to-outlet) have also been used to assess neuropathy [25]. Nevertheless, even these indices may be confounded by operator-induced and platform-specific sources of variability. Moreover, such morphometric measures lack information about the nerves' microstructure. Figure 2 shows B-mode images on the transverse plane (short axis plane) of the median nerve in an exemplary healthy participant compared with a CTS patient. Remarkably, the fascicles detected in the healthy median nerve cannot be seen in the CTS patient [5]. . The fascicles detected in the healthy median nerve cannot be seen in the CTS patient. This figure was previously presented by Vlassakov and Sala-Blanch [5]. Reprinting permission is granted through Rightslink system. Minor modifications were performed for presentation purposes.
Quantitative ultrasound (QUS) techniques offer the potential for a more objective evaluation of peripheral nerves, promising the improved diagnosis of neuropathy and nerve injury. QUS approaches have the potential to more reliably reflect differences in the multi-scale composite structure of the nerves (e.g., fascicles, organized bundles of nerve fibers within a fascicle, and axons), which cannot be directly observed by qualitative methods. QUS investigations performed on peripheral nerves can be categorized into three . The fascicles detected in the healthy median nerve cannot be seen in the CTS patient. This figure was previously presented by Vlassakov and Sala-Blanch [5]. Reprinting permission is granted through Rightslink system. Minor modifications were performed for presentation purposes.
Quantitative ultrasound (QUS) techniques offer the potential for a more objective evaluation of peripheral nerves, promising the improved diagnosis of neuropathy and nerve injury. QUS approaches have the potential to more reliably reflect differences in the multiscale composite structure of the nerves (e.g., fascicles, organized bundles of nerve fibers within a fascicle, and axons), which cannot be directly observed by qualitative methods. QUS investigations performed on peripheral nerves can be categorized into three main groups: (1) B-mode echogenicity-based outcomes, (2) elastography, and (3) backscattered RF signal characterization. B-mode echogenicity measures employ post-processing algorithms in B-mode images to determine the visualized microstructure of the tissue. The underlying physical principle of elastography is that tissue stiffness and other tissue mechanical properties can be quantitatively estimated by analyzing the response of the tissue to an applied force. Strain ultrasonography and shear wave elastography (SWE) are the two major classes of US elastography techniques [26]. In strain ultrasonography, tissue strain, as a response to internal or external compression stimuli, is measured by tracking the motion of speckles detectable in B-mode images [26,27]. In SWE, the speed of shear waves generated by external vibrations or tissue compression is measured and correlated to tissue elasticity or stiffness [26,27]. Finally, backscattered RF signal characterization permits the measurement and modeling of fundamental tissue acoustic parameters that relate to composition and microstructure [12,28]. The signals needed to measure these fundamental tissue parameters are often discarded during machine-dependent B-mode image formation.
In this review, we describe the three main categories of QUS techniques and summarize QUS-based investigations performed on peripheral nerves. The focus of this review will be the employed techniques rather than the results of each application case. In addition to summarizing the strengths and the use cases for each approach, this review also describes the potential limitations associated with each QUS technique.

Materials and Methods
This review conducted between 2020 and 2022 aimed to provide a thorough description of the current QUS techniques including the advantages and limitations. The literature search was performed in PubMed, Scopus, and Google Scholar databases using the following keywords: "peripheral nerve" and "quantitative ultrasound" or "elastography ultrasound" published after the year 1990. The results were first screened automatically and then through title and abstract reading for excluding duplicated records, non-English written reports, non-ultrasound or non-quantitative studies, and review articles. Study selection was performed following the PRISMA 2020 guidelines as summarized in Figure 3. All reviewed QUS-based investigations performed on peripheral nerves are summarized in Table 1.

B-Mode Echogenicity Measurement
In B-mode images, sonographers typically identify peripheral nerves based on fascicular boundaries; in the transverse plane, nerves appear honeycomb-like, while in the longitudinal plane, nerves display parallel hyperechoic fascicular borders. Figure 4 shows B-mode images in the transverse and longitudinal planes of the median nerve in a healthy volunteer generated with UFH22 (7-10 MHz) and UFH48 (10-22 MHz) US probes on a Vevo MD, FUJIFILM machine. The honeycomb-like structure of the nerves in the transverse plane becomes more obvious in images generated with higher-frequency probes. Nerves differ from tendons, which display a dense fibrillar structure in B-mode [7,69,70], and are additionally distinguished from blood vessels by using Doppler US [5][6][7]68]. B-mode echogenicity carries information about the nerve structure. Specifically, B-mode images comprise both the information of specular reflections at the tissue interfaces (large structures compared to the wavelength), displayed as hyperechoic areas, and the information of the backscattered signal from the microstructure of the tissue (equal to or smaller than the wavelength), displayed as speckle texture in the image. and are additionally distinguished from blood vessels by using Doppler US [5][6][7]68]. Bmode echogenicity carries information about the nerve structure. Specifically, B-mode images comprise both the information of specular reflections at the tissue interfaces (large structures compared to the wavelength), displayed as hyperechoic areas, and the information of the backscattered signal from the microstructure of the tissue (equal to or smaller than the wavelength), displayed as speckle texture in the image. The measurement of the hypoechoic fractional area of the nerves in US B-mode images, or so-called nerve density index, has been suggested as an objective measurement of overall nerve echogenicity, capable of distinguishing between normal and abnormal The measurement of the hypoechoic fractional area of the nerves in US B-mode images, or so-called nerve density index, has been suggested as an objective measurement of overall nerve echogenicity, capable of distinguishing between normal and abnormal nerves. Tagliafico et al. investigated the feasibility of using the peripheral nerve density index for distinguishing between CTS and neurofibromas in the median nerve [29]. Nerve density was found to be significantly higher in patients with CTS and significantly lower in patients with neurofibromas compared with healthy normal subjects. Using an analogous approach, Bignotti et al. concluded that the median nerve density index was significantly lower in patients with limited cutaneous systemic sclerosis compared to healthy controls [22]. Moreover, symptomatic patients demonstrated lower nerve density compared to nonsymptomatic patients. Such approaches have been further refined by normalizing nerve echogenicity to its surrounding tissue environment, thus potentially offsetting machine effects and absolute patient-to-patient differences in echogenicity. For example, Byra et al. investigated the nerve-tissue contrast index (NTI), the ratio between the average brightness of the surrounding tissue to the median nerve, in the context of a CTS diagnosis [24]. NTI was found to be significantly higher in CTS patients compared with healthy volunteers. Boom and Visser employed a variety of thresholding methods to measure the hypoechoic fraction (similar to US nerve density) of the ulnar nerve to detect ulnar neuropathy, with a significantly lower average hypoechoic fraction found in a patient group compared with controls [30]. In an attempt to correlate echogenicity measures with more conventional diagnostic approaches, Simon et al. [31] studied the relation between the ulnar nerve hypoechoic fraction and the electrophysiologic "inching" measures [31]. Hypoechoic fraction and CSA were significantly increased in patients with neuropathy immediately distal and proximal to the medial epicondyle. The above-elbow ulnar motor conduction velocity was inversely correlated with both CSA and hypoechoic fraction in limbs with neuropathy. As an example of the potential ambiguity of such approaches, though, asymptomatic and symptomatic limbs of patients demonstrated similar hypoechoic fractions. In addition to these more common echogenicity measures, the texture of US B-mode images may also be evaluated in peripheral nerves, offering a more nuanced assessment of structural heterogeneity within nerves. For example, in cadaveric ulnar nerves, a texture feature index obtained using the gray level co-occurrence matrix algorithm correlated to combined collagen and myelin concentrations obtained from histology [21].

US Elastography
The concept of US elastography was introduced by Ophir et al. [71], as a generalization of Eisenscher's echosonography method [72]. Elastography in general can be explained as analogous to a palpation exam performed on the entire tissue volume. During a palpation exam, the physician taps (shears) the tissue with his or her fingers, and qualitatively senses the deformability or stiffness of the examined tissues. Similarly, US elastography estimates tissue elasticity by analyzing the response of the tissue to an applied force monitored with US [26,27]. Strain ultrasonography and SWE are the two major classes of US elastography techniques [26]. In strain ultrasonography, the tissue strain, as a response to internal or external compression stimuli, is measured by tracking the motion of speckles detectable in B-mode images [26,27]. In SWE, the speed of shear waves induced by external vibrations or tissue compression is measured, which is correlated to tissue elasticity or stiffness [26,27]. These basic mechanical parameters are defined and presented schematically in Table 2. Table 2. Definition of the basic mechanical parameters described in this study.

Mechanical Parameters
Definition Formula Schematics

Normal stress
The magnitude of the force applied to the unit area perpendicular to the force direction  Table 2. Definition of the basic mechanical parameters described in this study.

Normal stress
The magnitude of the force applied to the unit area perpendicular to the force direction =

Normal strain
The change in length of the tissue per its unit length parallel to the force direction = ∆

Young's modulus (i.e., normal modulus)
Elasticity defined in applied force direction but perpendicular to the unit volume surface =

Shear stress
The magnitude of the force applied to the unit area parallel to the force direction =

Shear strain
The angular change in originally right angles of the unit volume after shear stress application = ∆

Shear modulus
Elasticity defined in applied force direction and parallel to the unit volume surface = = 2 1

Bulk modulus
Elasticity defined over the unit of volume when a uniform volumetric stress is applied, and a uniform strain is induced = 3 1 2 Elasticity and strain are two important concepts in strain elastography and SWE. The elasticity of a material describes its tendency to retain its original size and shape after being subjected to a deforming force or stress. Solids possess shear and volume elasticity such that they resist changes in shape and volume. Liquids, however, possess only volume elasticity such that they resist changes in their volume, but not in their shapes. Soft tissues are comprised of both liquids and solids; therefore, they possess shear and volume elasticities, yet their shear elasticity is significantly lower than their volume elasticity. Strain can be described as the change in the length of the tissue per its unit length ( Table  2). The magnitude of the force applied to the unit area, which induces the strain, is known as stress. The modulus of elasticity can be quantified as the ratio of stress to strain (units of N/m 2 , or Pa). It can be quantified depending on the stress and strain direction, such that Young's modulus (i.e., normal modulus), E, quantifies elasticity in the direction normal to the unit volume surface while the shear modulus, G, quantifies elasticity in the tangential direction to the unit volume surface. However, when uniform volumetric stress and strain can be assumed (e.g., liquids under pressure), the bulk or volume elastic modulus, K, can be described over the unit of volume.

Strain Ultrasonography
In strain ultrasonography, US is employed to measure the induced tissue displacement in the same direction as the applied stress. As shown in Figure 5, the required cyclic mechanical compression can be applied by (a) free-hand cyclic compression (palpation), (b) cardiovascular pulsation or respiratory motion, (c) acoustic radiation force impulse (ARFI), or (d) external mechanical vibration [2]. The induced strain, regardless of the stress application method, can be measured using different approaches depending on the

Normal strain
The change in length of the tissue per its unit length parallel to the force direction ε = ∆L  Table 2. Definition of the basic mechanical parameters described in this study.

Normal stress
The magnitude of the force applied to the unit area perpendicular to the force direction =

Normal strain
The change in length of the tissue per its unit length parallel to the force direction = ∆ Young's modulus (i.e., normal modulus) Elasticity defined in applied force direction but perpendicular to the unit volume surface =

Shear stress
The magnitude of the force applied to the unit area parallel to the force direction =

Shear strain
The angular change in originally right angles of the unit volume after shear stress application = ∆

Shear modulus
Elasticity defined in applied force direction and parallel to the unit volume surface = = 2 1

Bulk modulus
Elasticity defined over the unit of volume when a uniform volumetric stress is applied, and a uniform strain is induced = 3 1 2 Elasticity and strain are two important concepts in strain elastography and SWE. The elasticity of a material describes its tendency to retain its original size and shape after being subjected to a deforming force or stress. Solids possess shear and volume elasticity such that they resist changes in shape and volume. Liquids, however, possess only volume elasticity such that they resist changes in their volume, but not in their shapes. Soft tissues are comprised of both liquids and solids; therefore, they possess shear and volume elasticities, yet their shear elasticity is significantly lower than their volume elasticity. Strain can be described as the change in the length of the tissue per its unit length ( Table  2). The magnitude of the force applied to the unit area, which induces the strain, is known as stress. The modulus of elasticity can be quantified as the ratio of stress to strain (units of N/m 2 , or Pa). It can be quantified depending on the stress and strain direction, such that Young's modulus (i.e., normal modulus), E, quantifies elasticity in the direction normal to the unit volume surface while the shear modulus, G, quantifies elasticity in the tangential direction to the unit volume surface. However, when uniform volumetric stress and strain can be assumed (e.g., liquids under pressure), the bulk or volume elastic modulus, K, can be described over the unit of volume.

Strain Ultrasonography
In strain ultrasonography, US is employed to measure the induced tissue displacement in the same direction as the applied stress. As shown in Figure 5, the required cyclic mechanical compression can be applied by (a) free-hand cyclic compression (palpation), (b) cardiovascular pulsation or respiratory motion, (c) acoustic radiation force impulse (ARFI), or (d) external mechanical vibration [2]. The induced strain, regardless of the stress application method, can be measured using different approaches depending on the

Shear strain
The angular change in originally right angles of the unit volume after shear stress application

Shear modulus
Elasticity defined in applied force direction and parallel to the unit volume surface

Bulk modulus
Elasticity defined over the unit of volume when a uniform volumetric stress is applied, and a uniform strain is induced  Table 2. Definition of the basic mechanical parameters described in this study.

Normal stress
The magnitude of the force applied to the unit area perpendicular to the force direction =

Normal strain
The change in length of the tissue per its unit length parallel to the force direction = ∆

Young's modulus (i.e., normal modulus)
Elasticity defined in applied force direction but perpendicular to the unit volume surface =

Shear stress
The magnitude of the force applied to the unit area parallel to the force direction =

Shear strain
The angular change in originally right angles of the unit volume after shear stress application = ∆

Shear modulus
Elasticity defined in applied force direction and parallel to the unit volume surface = = 2 1

Bulk modulus
Elasticity defined over the unit of volume when a uniform volumetric stress is applied, and a uniform strain is induced = 3 1 2 Elasticity and strain are two important concepts in strain elastography and SWE. The elasticity of a material describes its tendency to retain its original size and shape after being subjected to a deforming force or stress. Solids possess shear and volume elasticity such that they resist changes in shape and volume. Liquids, however, possess only volume elasticity such that they resist changes in their volume, but not in their shapes. Soft tissues are comprised of both liquids and solids; therefore, they possess shear and volume elasticities, yet their shear elasticity is significantly lower than their volume elasticity. Strain can be described as the change in the length of the tissue per its unit length ( Table  2). The magnitude of the force applied to the unit area, which induces the strain, is known as stress. The modulus of elasticity can be quantified as the ratio of stress to strain (units of N/m 2 , or Pa). It can be quantified depending on the stress and strain direction, such that Young's modulus (i.e., normal modulus), E, quantifies elasticity in the direction normal to the unit volume surface while the shear modulus, G, quantifies elasticity in the tangential direction to the unit volume surface. However, when uniform volumetric stress and strain can be assumed (e.g., liquids under pressure), the bulk or volume elastic modulus, K, can be described over the unit of volume.

Strain Ultrasonography
In strain ultrasonography, US is employed to measure the induced tissue displacement in the same direction as the applied stress. As shown in Figure 5, the required cyclic mechanical compression can be applied by (a) free-hand cyclic compression (palpation), (b) cardiovascular pulsation or respiratory motion, (c) acoustic radiation force impulse (ARFI), or (d) external mechanical vibration [2]. The induced strain, regardless of the stress application method, can be measured using different approaches depending on the Elasticity and strain are two important concepts in strain elastography and SWE. The elasticity of a material describes its tendency to retain its original size and shape after being subjected to a deforming force or stress. Solids possess shear and volume elasticity such that they resist changes in shape and volume. Liquids, however, possess only volume elasticity such that they resist changes in their volume, but not in their shapes. Soft tissues are comprised of both liquids and solids; therefore, they possess shear and volume elasticities, yet their shear elasticity is significantly lower than their volume elasticity. Strain can be described as the change in the length of the tissue per its unit length ( Table 2). The magnitude of the force applied to the unit area, which induces the strain, is known as stress. The modulus of elasticity can be quantified as the ratio of stress to strain (units of N/m 2 , or Pa). It can be quantified depending on the stress and strain direction, such that Young's modulus (i.e., normal modulus), E, quantifies elasticity in the direction normal to the unit volume surface while the shear modulus, G, quantifies elasticity in the tangential direction to the unit volume surface. However, when uniform volumetric stress and strain can be assumed (e.g., liquids under pressure), the bulk or volume elastic modulus, K, can be described over the unit of volume.

Strain Ultrasonography
In strain ultrasonography, US is employed to measure the induced tissue displacement in the same direction as the applied stress. As shown in Figure 5, the required cyclic mechanical compression can be applied by (a) free-hand cyclic compression (palpation), (b) cardiovascular pulsation or respiratory motion, (c) acoustic radiation force impulse (ARFI), or (d) external mechanical vibration [2]. The induced strain, regardless of the stress application method, can be measured using different approaches depending on the manufacturer: RF echo correlation-based tracking, Doppler processing, or a combination of the two methods [2]. For example, in RF echo correlation-based tracking, RF A-lines are acquired along the axis of displacement; then, their changes in time between different acquisitions allow the measurement of the tissue displacement and the estimation of the normal strain. manufacturer: RF echo correlation-based tracking, Doppler processing, or a combination of the two methods [2]. For example, in RF echo correlation-based tracking, RF A-lines are acquired along the axis of displacement; then, their changes in time between different acquisitions allow the measurement of the tissue displacement and the estimation of the normal strain.
In the free-hand compression technique, compression-decompression cycles are performed using the US transducer itself, with force and frequency adjusted by the sonographer to an appropriate range according to a strain indicator on the US screen [34]. It is very important that the compression occurs vertically, without over-compression, and that the tissue does not slip out of the compression plane [33]. Strain ratios (SRs) are often calculated to provide more reproducible indices from strain-ultrasonography-generated images, as the applied stresses cannot be well-controlled [2]. In ARFI-based compression, a long-duration (e.g., 0.1-0.5 ms vs. 0.02 ms pulses in Bmode imaging), high-intensity (e.g., spatial peak pulse average = 1400 W/cm 2 , spatial peak temporal average = 0.7 W/cm 2 ) acoustic "pushing pulse" (i.e., ARFI) is used to displace tissue (displacement of ~10-20 μm) perpendicular to the surface. The magnitude of the applied acoustic radiation force, F, can be estimated by the acoustic absorption rate in the tissue, α, the speed of sound in the tissue, c, and the temporal average intensity of the acoustic beam, I, according to Equation (1) [2,26,27].

Strain Ultrasonography Applied to Peripheral Nerves
Orman et al. [33] investigated the application of free-hand compression in strain ultrasonography to evaluate median nerve mechanical properties. The mean tissue strain was significantly lower in patients with CTS compared to controls, implying higher tissue stiffness. Given that the mean median nerve perimeter and its CSA in patients were also significantly higher than those of controls, the authors concluded that the enlarged nerves were entrapped by the surrounding tissue structures in carpal tunnel. This entrapment created a pre-stressed condition, resulting in increased apparent stiffness; it was hypothesized that freeing the nerve would reduce the apparent stiffness. The reproducibility of the free-hand compression strain ultrasonography technique in detecting CTS was improved by calculating SR against an acoustic coupler rubber (ACR, made of elastic resin) as an external reference (Figure 6, [34]). The higher SR (ACR/nerve) in CTS patients compared with healthy volunteers implied lower nerve strain, similar to earlier studies [33]. Kesikburun et al. [39] later attempted using bone as an internal reference for SR In the free-hand compression technique, compression-decompression cycles are performed using the US transducer itself, with force and frequency adjusted by the sonographer to an appropriate range according to a strain indicator on the US screen [34]. It is very important that the compression occurs vertically, without over-compression, and that the tissue does not slip out of the compression plane [33]. Strain ratios (SRs) are often calculated to provide more reproducible indices from strain-ultrasonography-generated images, as the applied stresses cannot be well-controlled [2].
In ARFI-based compression, a long-duration (e.g., 0.1-0.5 ms vs. 0.02 ms pulses in Bmode imaging), high-intensity (e.g., spatial peak pulse average = 1400 W/cm 2 , spatial peak temporal average = 0.7 W/cm 2 ) acoustic "pushing pulse" (i.e., ARFI) is used to displace tissue (displacement of~10-20 µm) perpendicular to the surface. The magnitude of the applied acoustic radiation force, F, can be estimated by the acoustic absorption rate in the tissue, α, the speed of sound in the tissue, c, and the temporal average intensity of the acoustic beam, I, according to Equation (1) [2,26,27].

Strain Ultrasonography Applied to Peripheral Nerves
Orman et al. [33] investigated the application of free-hand compression in strain ultrasonography to evaluate median nerve mechanical properties. The mean tissue strain was significantly lower in patients with CTS compared to controls, implying higher tissue stiffness. Given that the mean median nerve perimeter and its CSA in patients were also significantly higher than those of controls, the authors concluded that the enlarged nerves were entrapped by the surrounding tissue structures in carpal tunnel. This entrapment created a pre-stressed condition, resulting in increased apparent stiffness; it was hypothesized that freeing the nerve would reduce the apparent stiffness. The reproducibility of the free-hand compression strain ultrasonography technique in detecting CTS was improved by calculating SR against an acoustic coupler rubber (ACR, made of elastic resin) as an external reference (Figure 6, [34]). The higher SR (ACR/nerve) in CTS patients compared with healthy volunteers implied lower nerve strain, similar to earlier studies [33]. Kesikburun et al. [39] later attempted using bone as an internal reference for SR calculation in median nerve evaluation in CTS patients, reporting a four-fold increase in nerve elasticity for the symptomatic wrist versus the asymptomatic wrist. However, considering bone as the reference for SR measurements is challenging, given the differences in sound speed within soft tissues versus bone. Moreover, near-zero values are expected for strains in bone using free-hand compression. In contrast, Nogueira-Barbosa et al. [41] used the flexor digitorum superficialis muscle (i.e., a soft tissue) as an internal reference for SR calculation in the median nerve in patients with chronic leprosy. A significantly lower SR was observed in the leprosy patients compared with the control group. Later, Tezcan et al. [43] used fat as the reference for SR measurement in the median nerve in patients with CTS undergoing low-level laser therapy and reported a significant drop in nerve stiffness (increased strain) after the therapy. Despite the promise of using muscle or fat as internal references, one caveat is that muscle, fat, and nerves may also be affected by a given pathological condition. calculation in median nerve evaluation in CTS patients, reporting a four-fold increase in nerve elasticity for the symptomatic wrist versus the asymptomatic wrist. However, considering bone as the reference for SR measurements is challenging, given the differences in sound speed within soft tissues versus bone. Moreover, near-zero values are expected for strains in bone using free-hand compression. In contrast, Nogueira-Barbosa et al. [41] used the flexor digitorum superficialis muscle (i.e., a soft tissue) as an internal reference for SR calculation in the median nerve in patients with chronic leprosy. A significantly lower SR was observed in the leprosy patients compared with the control group. Later, Tezcan et al. [43] used fat as the reference for SR measurement in the median nerve in patients with CTS undergoing low-level laser therapy and reported a significant drop in nerve stiffness (increased strain) after the therapy. Despite the promise of using muscle or fat as internal references, one caveat is that muscle, fat, and nerves may also be affected by a given pathological condition.  Figure 6b,d) with an acoustic coupler rubber (ACR) as an external reference (indicated with B in Figure 6b,d). The median nerve in the CTS patient showed a higher SR (ACR/Median nerve) and CSA. These figures were previously presented by Miyamoto et al. [34]. Reprinting permission is granted by the Radiological Society of North America, which is the copyright holder. Minor modifications were performed for presentation purposes. * = Ulnar artery, FCR = flexor carpi radialis, P = pisiform bone, S = scaphoid bone.
Free-hand compression strain ultrasonography was also used to probe neuropathic severity in several studies, including a comparatively subjective colorimetric assessment of strain levels [36] and a more repeatable SR-based approach [37], to distinguish between individuals with and without CTS. For the latter, the outcomes could not distinguish between mild and moderate CTS [37]. However, a comparable SR-based approach (reciprocal of SR) was successfully used to identify individuals with diabetic neuropathy in tibial nerves; here, the outcomes differentiated between neuropathic severity and correlated with morphometric changes [38].
Ambient or passive strain ultrasonography was also deployed to probe neuropathy; however, the measured SR between the median nerve and nearby tendons was insuffi-  B in (b,d)). The median nerve in the CTS patient showed a higher SR (ACR/Median nerve) and CSA. These figures were previously presented by Miyamoto et al. [34]. Reprinting permission is granted by the Radiological Society of North America, which is the copyright holder. Minor modifications were performed for presentation purposes. * = Ulnar artery, FCR = flexor carpi radialis, P = pisiform bone, S = scaphoid bone.
Free-hand compression strain ultrasonography was also used to probe neuropathic severity in several studies, including a comparatively subjective colorimetric assessment of strain levels [36] and a more repeatable SR-based approach [37], to distinguish between individuals with and without CTS. For the latter, the outcomes could not distinguish between mild and moderate CTS [37]. However, a comparable SR-based approach (reciprocal of SR) was successfully used to identify individuals with diabetic neuropathy in tibial nerves; here, the outcomes differentiated between neuropathic severity and correlated with morphometric changes [38].
Ambient or passive strain ultrasonography was also deployed to probe neuropathy; however, the measured SR between the median nerve and nearby tendons was insufficient to identify significant differences between CTS and control groups [42]. It is likely that cardiorespiratory pulsations induced inconsistent deformations in the nerve and reference tendons, which resulted in higher variability compared to free-hand compression strain ultrasonography. In contrast, strain ultrasonography with a cyclic compression apparatus improved repeatability by means of a pre-determined cyclic displacement of the transducer (4 mm displacement at 1.5 Hz) [35]. Using this approach, consistent with freehand compression studies, strains were significantly lower while SR and CSA were significantly higher in CTS patients versus controls. In a later study, the same research group [40] added a strain gauge to the cyclic compression apparatus, to measure pressure. The pressure and pressure/strain ratio (a more realistic index to determine elasticity) were both significantly higher in the CTS patients than in controls.
The feasibility of strain ultrasonography using ARFI push pulses was first investigated by Palmeri et al. [32] to visualize peripheral nerves with adequate contrast versus their surrounding tissues (e.g., muscle, fat, and fascia). The purpose was to improve US guidance in monitoring the distribution of injected anesthetic around the targeted nerves. ARFI strain ultrasonography images yielded significant contrast improvements for the distal sciatic nerve structures and brachial plexus peripheral nerves compared with B-mode imaging. B-mode and ARFI image acquisitions were ECG-triggered for in vivo imaging at locations adjacent to arteries. Several ARFI-based strain ultrasonography investigations have also been reported in the literature as parts of SWE-focused studies, which are summarized in the next section. A limitation of ARFI-based approaches is that the intensity of the radiation force is limited by the potential tissue damage by a push at high acoustic power. As a consequence, the magnitude of the displacements as well as the imaging depth are restricted. In the next section, the assessment of shear elastic modulus (G) will be discussed, as it has more recently been preferred for its fast and reliable assessment, and its wider range of measurable values.

Shear Wave Elastography (SWE)
In contrast to strain ultrasonography, which measures physical tissue displacement parallel to the applied normal stress, SWE employs dynamic stress to generate shear waves in perpendicular directions. The measurement of the shear wave speed (SWS) results in an estimation of the tissue's shear and normal elastic moduli (G and E) [27]. SWE was developed based on the fact that the shear elastic modulus, G (G ≈ E/3 for semiincompressible tissues), determines the propagation speed of mechanical waves, SWS, at a magnitude of the square root of G divided by tissue density, ρ, as presented in Equation (2).
Thus, by evaluating SWS, the elastic modulus of a medium can be estimated. SWS remains below 10 m/s on average in soft tissues and can be adequately tracked by B-mode US images [27]. Shear waves propagate faster through stiffer and denser tissues, as well as along the tissue axis aligned with organized fibers (e.g., the long axis in tendons) [73]. As shown in Figure 7, the shear wave generation methods in nerve SWE studies can be categorized mainly as (a) external mechanical vibration, (b) single-point-focused ARFI, and (c) multiple-point-focused ARFI [2]. Although an external vibration device can induce shear waves, as initially proposed by Krouskop et al. [74], most clinical SWE studies utilize ARFI pulses to induce shear waves, as first proposed by Sugimoto et al. in 1990 [75]. ARFI pushing pulses perpendicular to the tissue surface, as noted in Section 3, result in tissue vibration at an ultrasonic frequency and consequent tissue deformation (displacement) within the region of US excitation (ROE), due to sound wave absorption and scattering. The induced deformation generates shear waves that laterally propagate away from the ROE at a much slower velocity (<10 m/s) compared with US pressure pulses (1500 m/s). Although shear wave mode conversion also occurs with conventional B-mode imaging, the force magnitudes are too small to generate tissue motion detectable with conventional US. ARFI pulses can be applied at a single focal location (point shear wave elastography, pSWE) ( Figure 7b) or a multi-focal configuration such that each focal zone is interrogated in rapid succession, leading to a cylindrically shaped shear wave extending over a larger depth, enabling realtime shear wave images to be formed ( Figure 7c) [2,26,27]. SWE with multi-focal ARFI configuration is called 2D shear wave elastography (2D-SWE) and allows the real-time monitoring of shear waves in 2D for an SWS measurement. Notably, for 2D SWE, shorter propagation distances are utilized due to the limitations in the number of US tracking pulses.
Considering shear wave generation as the first step in SWE, during the second step, US rapid plane wave excitations are used to track tissue displacement while shear waves propagate. In the third step, changes in tissue displacement maps over time are used to calculate SWS. Frame rates for tracking shear waves are typically between 2 kHz and 10 kHz [73]. Scanners may display different versions of a quality index as a measure of confidence in the estimated SWS, which is calculated from the correlation coefficients between frames of the speckle-tracking images. If the frame rate is too low, there is motion of the patient or probe, or there is no well-developed speckle in the region, then the quality index is low.

SWE Applied to Peripheral Nerves
ARFI-SWE has been investigated in several studies to detect CTS. The median nerve stiffness and SWS were reported to be significantly higher in CTS patients compared with controls, and often accompanied by an increase in nerve CSA [44,50,52,64]. Stiffness also accurately differentiated between patients with severe CTS and those with mild or moderate [44,50,52,64]. Figure 8 shows the B-mode image, estimated stiffness in kPa, and the shear wave propagation map in the median nerves of a healthy participant (a, b) and a CTS patient (c, d) [52], who demonstrated a higher stiffness and larger CSA. Notably, the shear wave propagation map displays the arrival time of a shear wave using a series of Although an external vibration device can induce shear waves, as initially proposed by Krouskop et al. [74], most clinical SWE studies utilize ARFI pulses to induce shear waves, as first proposed by Sugimoto et al. in 1990 [75]. ARFI pushing pulses perpendicular to the tissue surface, as noted in Section 3, result in tissue vibration at an ultrasonic frequency and consequent tissue deformation (displacement) within the region of US excitation (ROE), due to sound wave absorption and scattering. The induced deformation generates shear waves that laterally propagate away from the ROE at a much slower velocity (<10 m/s) compared with US pressure pulses (1500 m/s). Although shear wave mode conversion also occurs with conventional B-mode imaging, the force magnitudes are too small to generate tissue motion detectable with conventional US. ARFI pulses can be applied at a single focal location (point shear wave elastography, pSWE) ( Figure 7b) or a multi-focal configuration such that each focal zone is interrogated in rapid succession, leading to a cylindrically shaped shear wave extending over a larger depth, enabling real-time shear wave images to be formed ( Figure 7c) [2,26,27]. SWE with multi-focal ARFI configuration is called 2D shear wave elastography (2D-SWE) and allows the real-time monitoring of shear waves in 2D for an SWS measurement. Notably, for 2D SWE, shorter propagation distances are utilized due to the limitations in the number of US tracking pulses.
Considering shear wave generation as the first step in SWE, during the second step, US rapid plane wave excitations are used to track tissue displacement while shear waves propagate. In the third step, changes in tissue displacement maps over time are used to calculate SWS. Frame rates for tracking shear waves are typically between 2 kHz and 10 kHz [73]. Scanners may display different versions of a quality index as a measure of confidence in the estimated SWS, which is calculated from the correlation coefficients between frames of the speckle-tracking images. If the frame rate is too low, there is motion of the patient or probe, or there is no well-developed speckle in the region, then the quality index is low.

SWE Applied to Peripheral Nerves
ARFI-SWE has been investigated in several studies to detect CTS. The median nerve stiffness and SWS were reported to be significantly higher in CTS patients compared with controls, and often accompanied by an increase in nerve CSA [44,50,52,64]. Stiffness also accurately differentiated between patients with severe CTS and those with mild or moderate [44,50,52,64]. Figure 8 shows the B-mode image, estimated stiffness in kPa, and the shear wave propagation map in the median nerves of a healthy participant (a, b) and a CTS patient (c, d) [52], who demonstrated a higher stiffness and larger CSA. Notably, the shear wave propagation map displays the arrival time of a shear wave using a series of wave contour lines. ARFI-based SWE also revealed similar changes in the median nerve (higher nerve stiffness and CSA) for patients with acromegaly [57]. wave contour lines. ARFI-based SWE also revealed similar changes in the median nerve (higher nerve stiffness and CSA) for patients with acromegaly [57]. Several studies have focused on the SWE-based assessment of peripheral nerve neuropathy. Aslan and Analan [54] used SWE to study the effects of chronic flexed wrist posture among chronic stroke patients on median nerve elasticity and CSA. SWS was significantly higher, but CSA was significantly lower (unlike CTS) for the affected side than the unaffected side; however, SWS did not correlate with the elapsed time since stroke. Similarly, SWE revealed significantly greater ulnar nerve stiffness within the cubital tunnel in patients with ulnar neuropathy (i.e., cubital tunnel syndrome) compared to healthy controls [56]; tibial nerve stiffness in patients with diabetic neuropathy compared to non-neuropathic diabetics and healthy controls [48,60,62,65]; and median nerve stiffness in patients with diabetic neuropathy compared to controls [61,62]. Neuropathic changes were often accompanied by increased CSA [60][61][62]; however, some studies have reported a reversed trend in the tibial nerve [65]. Interestingly, though, both the median and tibial nerves were often found to be thinner in diabetic patients who did not display neuropathy, suggestive of nerve atrophy before neuropathy onset [62]. Conversely, other studies have reported higher CSAs in the tibial nerves of patients with diabetic neuropathy [65]. Such controversial findings in CSA changes emphasize the utility of combining QUS and morphometric imaging.
A number of other nerves have also been studied using SWE. SWE and free-hand compression strain ultrasonography in optic nerves of patients with Behcet's syndrome (a chronic autoimmune disease) [46] and multiple sclerosis [47] showed significantly higher stiffness and lower strains compared with the control group. Optic nerve stiffness was also examined in patients with migraine using SWE and free-hand compression strain ultrasonography; patients displayed an insignificant decrease in calculated strain (stiffness increase), but a significant increase in SWS [63]. In another study, SWE revealed higher elastic moduli in the brachial plexus nerves of patients receiving radiotherapy compared to their contralateral nerves [55]. SWE has also been used to evaluate healthy nerves; for example, sex-dependent differences between elastic moduli in cervical nerve roots have been noted in healthy volunteers [53]. Several studies have focused on the SWE-based assessment of peripheral nerve neuropathy. Aslan and Analan [54] used SWE to study the effects of chronic flexed wrist posture among chronic stroke patients on median nerve elasticity and CSA. SWS was significantly higher, but CSA was significantly lower (unlike CTS) for the affected side than the unaffected side; however, SWS did not correlate with the elapsed time since stroke. Similarly, SWE revealed significantly greater ulnar nerve stiffness within the cubital tunnel in patients with ulnar neuropathy (i.e., cubital tunnel syndrome) compared to healthy controls [56]; tibial nerve stiffness in patients with diabetic neuropathy compared to non-neuropathic diabetics and healthy controls [48,60,62,65]; and median nerve stiffness in patients with diabetic neuropathy compared to controls [61,62]. Neuropathic changes were often accompanied by increased CSA [60][61][62]; however, some studies have reported a reversed trend in the tibial nerve [65]. Interestingly, though, both the median and tibial nerves were often found to be thinner in diabetic patients who did not display neuropathy, suggestive of nerve atrophy before neuropathy onset [62]. Conversely, other studies have reported higher CSAs in the tibial nerves of patients with diabetic neuropathy [65]. Such controversial findings in CSA changes emphasize the utility of combining QUS and morphometric imaging.
A number of other nerves have also been studied using SWE. SWE and free-hand compression strain ultrasonography in optic nerves of patients with Behcet's syndrome (a chronic autoimmune disease) [46] and multiple sclerosis [47] showed significantly higher stiffness and lower strains compared with the control group. Optic nerve stiffness was also examined in patients with migraine using SWE and free-hand compression strain ultrasonography; patients displayed an insignificant decrease in calculated strain (stiffness increase), but a significant increase in SWS [63]. In another study, SWE revealed higher elastic moduli in the brachial plexus nerves of patients receiving radiotherapy compared to their contralateral nerves [55]. SWE has also been used to evaluate healthy nerves; for example, sex-dependent differences between elastic moduli in cervical nerve roots have been noted in healthy volunteers [53].

Considerations for Evaluating Peripheral Nerves with SWE
Despite its broad utility, the scale of a targeted nerve should be carefully considered when interpreting SWE outcomes. In one study, the mean stiffness of the saphenous nerve was found to be very similar in the short and long axes [66], a finding inconsistent with multiple other investigations showing a higher shear wave speed in the longitudinal plane [1,67]. It is likely that reliable elasticity data could not be acquired in the transverse plane of the nerves [51], potentially due to the similar scale of nerve diameter and the wavelength of the shear waves and a higher signal from wave reflections at the fiber bundles, apparent in the nerve's transverse plane [1,67]. SWE measures in thinner nerves may also be less accurate as they are spanned by a smaller number of shear wavelengths to be tracked by US. In addition, the speckle tracking is confounded by the fact that the US beam is larger than the nerve, which introduces a partial volume effect that creates additional influences of the surrounding tissues on reported SWS.
Nerve anatomy and tension should also be considered in evaluating SWE outcomes. Median nerves stretched via wrist extension showed a significant increase in SWS (i.e., nerve stiffness) at the wrist compared to mid-forearm [51], possibly due to the (carpal) bone-proximity artifact [49], which occurs when the studied structure is near a rigid plane such as the bone cortex that prevents homogeneous shear wave propagation. Alternatively, increased SWS may reflect nerve tension due to nerve deformation over the extended wrist [76] or increased pressure within the carpal tunnel [77]. Similar phenomena were observed in sciatic nerves, where increased SWS was noted after tensioning through ankle dorsiflexion in combination with knee extension [45], in median nerves with elbow extension [68], and in ulnar nerves with elbow flexion [68].
The correlation between increased tensile loading and increased SWS has also been validated through the ex vivo application of tensile loads [51]. Nerve viability may also influence SWE outcomes; unlike the above study [51], Schrier et al. [67] reported no differences in median nerve elasticity between the median nerve at the wrist and forearm, as measured in cadaveric nerves using SWE and lateral mechanical compression tests. Such differences might be due to the loss of in vivo loading and intra-nerve pressure or postmortem tissue changes. More recently, Neto et al. [58] investigated the immediate impact of the slump neurodynamics technique, an exercise for the purpose of improving neural mechanosensitivity, on sciatic nerve stiffness using SWE. SWS was measured during passive foot dorsiflexion in a healthy group before and immediately after slump neurodynamics training. Despite significant differences in the sciatic nerve SWS caused by dorsiflexion, SWS did not change after slump neurodynamic loading, suggesting that nerves likely returned to their natural biomechanical state immediately after releasing the tensile load imposed by the slump. The impact of chronic lower-back-related leg pain on sciatic nerve stiffness was investigated during passive foot dorsiflexion by the same research group [59]. The affected limb showed higher sciatic nerve stiffness compared to the unaffected limb of patients; however, no differences were observed between the unaffected limb of patients versus the healthy controls.

Backscattered Radiofrequency (RF) Signal Characterization
US scanners increasingly permit access to the beam-formed backscattered RF signals through a research interface. Quantifying such raw RF US data allows an assessment of fundamental acoustic tissue parameters, which can be scanner-independent [12,28]. Commonly used RF signal quantifications, such as backscatter coefficient (BSC), attenuation coefficient (AC), structure-function, and stochastic modeling of envelope statistics of backscattered echoes, are highly likely to be related to the tissue composition and microstructure [12,28].
BSC indicates the tissue's ability to scatter US waves back to the transducer analogous to echogenicity, providing information about tissue structure and composition. Indeed, the backscattered coefficient represents the energy backscattered by the biological tissue as a fraction of the emitted signal. A reliable assessment requires accounting for all the causes of the loss of energy of the emitted wave in addition to the backscatter, such as the quantitative assessment of the tissue AC, but also the diffraction of the transducer and settings on the gain and processing in the scanner. AC indicates the magnitude of the loss of energy (absorption and scattering) during the US wave propagation into and back from tissues. The entropy of the backscattered echo amplitude can also be used as a model-free approach to analyze the backscattered echo statistics [78]. Such model-free approaches in general do not directly provide an estimation of the tissue microstructure, even though statistical correlations can be found between such coefficients and the tissue microstructural parameters.
In addition to model-free approaches to analyze backscattered echo statistics, some model-based methods have been proposed to estimate the microstructural properties of the tissues [12,79]. Such techniques often require solving an inverse equation problem by fitting the accurately measured BSC and AC parameters (functions of the RF spectrum) in a predefined model of the scatterers in the tissue of interest. For example, stochastic models of the backscattered echo statistics provide an indirect estimate of the spatial and size distributions of scattering microstructures within the studied tissue. Envelope statistics can be also modeled using various distribution functions such as Nakagami distribution, which models random distributions of scatterers in the resolution cells (pre-Rayleigh to Rician distributions) [80].
Raw backscattered RF signal characteristics have been used for several soft tissue assessments [12] including liver [81], breast [82], muscle [83], annular pulleys (tendons) of the fingers [84], and skin [85]. Recently, Byra et al. [21] investigated cadaveric ulnar nerves using raw RF signal measures including BSC, AC, Nakagami parameter, and entropy. These specimens were also assessed with histology; the combined collagen/myelin content demonstrated significant correlations with the BSC and entropy.

QUS Limitations in Nerve Evaluation
Employing any US technique for peripheral nerve assessment is affected by welldescribed sonography artifacts such as shadowing, insonation angle, reverberation, and clutter artifacts. Moreover, US techniques are affected by variations in the system settings and parameters such as RF frequency, sampling rate, and gains, which occasionally lead to biased results. Since the structures of interest in peripheral nerves are small, higher US frequencies are often preferred for qualitative and quantitative nerve imaging. The higher US frequencies result in a sharp decrease in the signal-to-noise ratio (SNR) as a function of the depth, due to the higher capacity of biological tissues to attenuate high-frequency RF waves. Therefore, the effective imaging depth for quantitative US techniques is often limited to a few centimeters for nerve imaging, which is more challenging for in vivo investigations when considering the variability in the dimensions of the intermediate tissue layers (e.g., skin, muscle, and fat) between the transducer and the nerves.
B-mode echogenicity measures performed on post-processed images (e.g., nerve density) are limited due to their dependence on the scanner settings and manual identification of tissue boundaries [1,2,73].
US elastography-based measures of tissues are limited by assumptions about the tissue's material behavior (e.g., linear, elastic, isotropic, homogenous, and incompressible material) to simplify the analysis and interpretation of the strain ultrasonography and SWE measurements [2,73]. However, soft tissues are known to be inherently nonlinear, viscoelastic, and heterogeneous [2]. Specifically, including viscosity in the tissue model means that the tissue stiffness and SWS both depend on the excitation ARFI pulse frequency (or frequency of the external mechanical vibrations), which may vary for different US systems and transducers [2]. Moreover, the mechanical nonlinearity in soft tissue's behavior means that the induced level of strain in response to an ARFI pulse or other external loads depends on the initial strain state of the material; in other words, operator-dependent transducer orientation and initial compression [1,2]. SWE and strain ultrasonography studies of peripheral nerves have recommended single cutoff values for strain, strain ratio, stiffness, and stiffness ratio between different nerve sections, which, in general, showed better sensitivity and specificity levels in neuropathy diagnosis compared with CSA values and their ratio, despite the codependence of CSA and SWS [1]. However, such single cutoff values show large variations depending on the US system and transducer [1].
Backscattered RF signal characterizations often require analyzing the tissue scans relative to a tissue-mimicking calibration phantom, with a known BSC and AC profile in a range of targeting frequencies. These extra calibration scans may be considered highly resource-or time-consuming, and require specialized training. Therefore, backscattered RF signal characterization methods may be especially challenging for patient-oriented in vivo experimental acquisitions. Notably, methods without extra calibration scans have been reported in other tissues (liver and breast) and may serve to guide future peripheral nerve assessments. In such methods, a smaller reference phantom can be scanned concurrently with the tissue and utilized for further BSC corrections. On the other hand, clinical US systems have been shown to have very stable acoustic output and gain, suggesting calibration scans can be stored once and then applied to the subsequent data analysis for extended periods, reducing the impact on clinical throughput. Moreover, the backscattered RF signal characterizations involve more sophisticated data processing, perhaps offline, compared with other QUS categories. As described in Section 5, the model-based backscatteredsignal-related measures often require solving an inverse equation problem by fitting the accurately measured BSC and AC parameters in a predefined model of the scatterers in the medium of interest. More sophisticated models may be required for a more accurate assessment of the backscattered RF signal from peripheral nerves by considering their specific microstructure, surrounding tissue, and microenvironment as quantified using immunohistochemical approaches [86]. As is the case for other QUS techniques, raw RF signal characterization can be influenced by the transducer orientation and compression applied by the operator.

Conclusions
Three main classes of QUS techniques were described and their reported applications on peripheral nerves in the literature were summarized in this study. Neuropathy and nerve injury may affect the composite structure of the nerves at different scales (i.e., fascicles, organized bundles of nerve fibers within a fascicle, and axons), which can be potentially assessed by QUS techniques. Among the discussed QUS techniques, SWE is more appropriate for the mechanical assessment of peripheral nerves while backscattered RF signal characterization is recommended for the microstructural assessment of the nerves. These approaches are important topics for further investigation.