Human skeletal muscle size with ultrasound imaging: a comprehensive review

Abstract

Skeletal muscle size is an important factor in assessing adaptation to exercise training and detraining, athletic performance, age-associated atrophy and mobility decline, clinical conditions associated with cachexia, and overall skeletal muscle health. Magnetic resonance (MR) imaging and computed tomography (CT) are widely accepted as the gold standard methods for skeletal muscle size quantification. However, it is not always feasible to use these methods (e.g., field studies, bedside studies, and large cohort studies). Ultrasound has been available for skeletal muscle examination for more than 50 years and the development, utility, and validity of ultrasound imaging are underappreciated. It is now possible to use ultrasound in situations where MR and CT imaging are not suitable. This review provides a comprehensive summary of ultrasound imaging and human skeletal muscle size assessment. Since the first study in 1968, more than 600 articles have used ultrasound to examine the cross-sectional area and/or volume of 107 different skeletal muscles in more than 27,500 subjects of various ages, health status, and fitness conditions. Data from these studies, supported by decades of technological developments, collectively show that ultrasonography is a valid tool for skeletal muscle size quantification. Considering the wide-ranging connections between human health and function and skeletal muscle mass, the utility of ultrasound imaging will allow it to be employed in research investigations and clinical practice in ways not previously appreciated or considered.

INTRODUCTION

Skeletal muscle mass measurement through whole muscle imaging allows for the quantitative assessment of muscular health and adaptations in various clinical and research scenarios. In 1968, Ikai and Fukunaga (1) were the first to use ultrasound imaging to assess muscle cross-sectional area (CSA) and examined the arm flexor muscle with a 2.25–5 MHz scanner. The same group also conducted an exercise training study and reported an increase in the CSA after 40 and 100 days of resistance exercise training (2). However, ultrasound imaging for quantitative assessment of skeletal muscle did not receive much attention at the time, mainly due to the relatively low image resolution limiting muscle differentiation, as well as the original technique required the limb to be submerged in a water tank. Before ultrasonography overcame these difficulties, significant developments in other imaging technologies emerged. In 1978, 6 years after a commercial computed tomography (CT) scanner was first introduced, thigh skeletal muscle CSA was assessed with a CT scan (3). Ten years later in 1988, quantitative assessments of back and lower extremity muscles were conducted with magnetic resonance imaging (MRI) (4, 5). MR and CT imaging are now firmly established as the gold standard for muscle specific assessment of skeletal muscle size. These techniques have been widely used by us and others to quantify muscle size in studies examining aging (6–11), exercise training (12–17), astronauts and microgravity simulations (18–25), athletes (26–30), and numerous clinical and nonclinical research populations (31–38). However, employing these methods is not always possible in some research studies or certain aspects of clinical practice for various reasons, including cost, clinical demand on the scanners, or proximity.

Brightness mode (B-mode) ultrasound imaging has been extensively developed and widely used to study human skeletal muscle structure in a wide range of conditions of aging, health, and fitness status. This technique allows a reproducible assessment of skeletal muscle with a relatively low cost, short execution time, and portability. After five decades of developments and a large number of investigations, the validity and role of ultrasound for skeletal muscle mass assessment can be reviewed for its research and clinical application. There are several reviews discussing ultrasound assessment of human skeletal muscle, yet these reviews tend to be region-specific (39–41), age-specific (42–48), disease-specific (41, 45, 48–52), technique-specific (53), or include a relatively small number of studies (54). Therefore, the goal of this review was to provide an up to date, comprehensive presentation of the existing literature of ultrasound imaging assessment of human skeletal muscle size.

SUMMARY OF ULTRASOUND LITERATURE

A comprehensive search was conducted for articles indexed on PubMed on or before December 31, 2021. The search term structure was determined based on SARCopenia through UltraSound (SARCUS) working group (43, 44), with modifications to eliminate aging-related terms to encompass all age groups and to add muscle size-related terms. Therefore, the following search term structure was used for this review: ((((((((Ultrasonography[Mesh]) OR Ultrasound) OR Echograph*) OR Ultrasonograph*) OR Ultrasonic) OR Echotomograph*) OR Sonograph*)) AND (“Muscles”[Mesh] OR “Lean tissue” OR “Lean mass” OR “Lean body mass” OR Muscle OR “Fat free mass”) AND (“Cross-sectional area” OR “Muscle volume”).

All articles were screened for eligibility regarding our inclusion and exclusion criteria and were included if they measured the size (anatomical CSA or volume) of voluntary skeletal muscles in humans using ultrasonography. Studies of cadavers and fetuses were also included. Articles referenced in the screened articles, but not appearing in the search results, were included if they measured muscle size in humans with ultrasound. Non-English language articles were excluded. Articles that estimated skeletal muscle CSA or volume based on muscle architecture (i.e., muscle thickness) were excluded, as MR and CT imaging almost exclusively assess muscle CSA and volume (through multiple CSA measurements) rather than thickness. The literature search yielded 2,726 articles in PubMed. In total, 627 articles were included in this review (Fig. 1A). The list of these articles is available in Supplemental Table S1 (see https://doi.org/10.6084/m9.figshare.19004915.v1) and Supplemental Table S2 (see https://doi.org/10.6084/m9.figshare.19003088.v1).

Figure 1.

Summary of human skeletal muscle size investigations with ultrasound. Articles published on or before December 31, 2021 were included. A: number of articles per year. B: number of subjects per year. C: subject characteristics distribution.

In total, the skeletal muscle size of 27,576 subjects was assessed with ultrasound from 1968 to December 31, 2021 (Fig. 1B). Subject characteristics were categorized as Healthy, Disease, Intensive Care Unit (ICU), Athlete, Healthy + Disease, Healthy + Athlete, and Cadaver. Over 50% of the subjects (>14,500) were healthy individuals. Approximately 30% of the subjects were diagnosed as having a disease (including benign diseases, cancers, cardiovascular diseases, digestive disorders, genetic disorders, immune system disorders, infectious diseases, inflammatory diseases, joint disorders, metabolic disorders, muscular disorders, neurological disorders, neuromuscular diseases, obese and overweight, renal disorders, respiratory diseases, and urinary disorders). In addition to these disease conditions, bedside muscle size assessments of over 1,600 critically ill patients admitted to an ICU facility for treatment were conducted with ultrasonography. Subjects were categorized as Athlete if subjects played a sport professionally or interscholastically, or if they were referred to as athletes or elite. Over 2,300 athletes (including alpine skiing, American football, archery, badminton, ballet dancing, ballroom dancing, baseball, basketball, bodybuilding, classic dancing, climbing, contemporary dancing, cricket, cross-country running, cross-country skiing, crossfit, cycling, decathletes, distance running, diving, fencing, gymnastics, handball, heptathletes, high jump, hurdle, ice hockey, javelin, judo, jumping, lifesaving, orienteering, pole vault, powerlifting, rowing, rugby, ski jumping, soccer, speed skate, swimming, tennis, throwing, track running, volleyball, water polo, weightlifting, and wrestling) underwent skeletal muscle assessments with ultrasound. Cross-sectional studies comparing diseased or athlete subjects to healthy subjects were classified as Healthy + Disease or Healthy + Athlete. In addition, over 60 cadavers were utilized for ultrasound skeletal muscle assessment. Characteristics of the subjects are summarized in Fig. 1C.

Overall, the size of 107 different skeletal muscles has been examined with ultrasonography (Table 1 and Supplemental Table S1). Muscles of the upper leg were the most frequently assessed (Fig. 2). Rectus femoris (207 articles), vastus lateralis (168 articles), gastrocnemius (73 articles), and lumbar multifidus (64 articles) were frequently examined for size with ultrasound imaging. The rectus femoris is the most commonly investigated muscle with ultrasonography likely because 1) the rectus femoris is easy to identify, 2) the analysis of the rectus femoris in most individuals can be done with a single image, and 3) the rectus femoris is a part of the quadriceps femoris, which is considered a functionally important muscle for the performance of daily living (55). Further, an assessment of one of the quadriceps muscles can be advantageous in various clinical situations. In fact, most of the studies (32 of 33) that imaged skeletal muscles of patients in ICU assessed the rectus femoris for muscle size quantification (56–87). Furthermore, only 6 of the 100 articles that solely measured rectus femoris used image construction methods (88–93), which will be discussed in ultrasound technology developments in skeletal muscle imaging. These data suggest that rectus femoris assessment with ultrasound is a relatively easy, quick, and cost-efficient method for examining human skeletal muscle health. However, the size of rectus femoris only constitutes ∼10% of the total quadriceps femoris in young and old individuals (11). Thus, the vastus lateralis, a quadriceps muscle that is also the most commonly biopsied muscle for skeletal muscle research studies (94–98), is frequently assessed. Image construction is required to assess vastus lateralis size with real-time B-mode ultrasound scanners.

Table 1.

Skeletal muscles studied in the articles included in this review

Muscle Group	Muscles
Face	Depressor anguli oris, depressor labii inferioris, digastric, diglossus, extraocular muscles, frontalis, geniohyoid, masseter, mentalis, mylohyoid, orbicularis oculli, orbicularis oris, temporalis, tongue (or genioglossus), zygomaticus major
Neck	Cervical multifidus, longissimus capitis, longus capitis, longus colli, obliquus capitis inferior, scalenus anterior, semispinalis capitis, semispinalis cervicis, short rotators, splenius capitis, splenius cervicis, sternocleidomastoid
Shoulder	Infraspinatus, levator scapulae, supraspinatus, trapezius
Trunk	External oblique, iliocostalis, iliopsoas, internal oblique, longissimus dorsi, lumbar multifidus, pectoralis major, quadratus lumborum, rectus abdominis, semispinalis, thoracic multifidus, transverse abdominis
Upper arm	Biceps brachii, brachialis, transferred gracilis, triceps brachii
Forearm	Abductor pollicis longus, brachioradialis, extensor carpi radialis brevis, extensor carpi radialis longus, extensor carpi ulnaris, extensor digiti minimi, extensor digitorum communis, extensor indicis proprius, flexor carpi radialis, flexor carpi ulnaris, flexor digitorum profundus, flexor digitorum superficialis, palmaris longus, pronator teres, supinator
Hand	Abductor digiti minimi of the hand, abductor pollicis brevis, abductor pollicis longus, first dorsal interosseous, lumbricals, opponens pollicis, pronator quadratus
Pelvic	Bulbospongiosus, external sphincter, ischiocavernosus, levator ani, levator hiatus, piriformis, rhabdomyosphincter, urethra surrounded striated muscles
Upper leg	Adductor magnus, biceps femoris, gracilis, quadratus femoris, rectus femoris, sartorius, semimembranosus, semitendinosus, tensor of vastus intermedius, vastus intermedius, vastus lateralis, vastus medialis
Lower leg	Extensor digitorum longus, extensor hallucis longus, flexor digitorum longus, flexor hallucis longus, lateral gastrocnemius, medial gastrocnemius, peroneus brevis, peroneus longus, peroneus tertius, soleus, tibialis anterior, tibialis posterior
Foot	Abductor digiti minimi of the foot, abductor hallucis, extensor digitorum brevis, flexor digitorum brevis, flexor hallucis brevis, quadratus plantae

Figure 2.

Summary of ultrasound assessment of human skeletal muscle cross-sectional area and volume. Body part distribution corresponds to the following muscle groups in Table 1: upper leg (upper leg), lower leg & foot (lower leg, foot), neck & trunk (neck, shoulder, trunk), upper extremity (upper arm, forearm, hand), others (face, pelvic). Numbers in the chart represent the number of articles that examined the body parts and muscles in the upper leg. Some articles examined multiple body parts or multiple muscles in the upper leg; thus, the sums do not correspond to the total number of articles and the total number of upper leg investigations.

Gastrocnemius is also an important muscle for postural balance (99) and specific responses to aging (6, 100) and unloading (23, 25) have been investigated. Ultrasound assessment of this calf muscle also requires image construction to examine the whole muscle size. The medial head of the gastrocnemius is more commonly examined with ultrasound imaging and it is generally larger than its lateral head (101, 102).

Lumbar multifidus is generally the largest among all multifidus muscles that run along the whole vertebral column. Although the function of lumbar multifidus is not commonly examined due to the difficulty in performing an isolated movement, it plays a role in stabilizing the lumbar spine and its atrophy has been shown to be accompanied by chronic low back pain (103, 104). It is also beneficial that the lumbar multifidus can be captured in a single ultrasound image.

The distribution of ultrasound brands used for each study is summarized in Fig. 3. Ultrasound systems from GE Healthcare were most frequently used in muscle size assessment. Of those, a portable LOGIQ e ultrasound imaging system was commonly used with a 12 L-RS linear probe (5–13 MHz, 38.4 mm footprint). Aloka ultrasound systems dominated the early years of skeletal muscle examination and was the primary brand choice until 2017. A portable M-Turbo ultrasound system from Fujifilm SonoSite has been most frequently used in ICU bedside muscle assessments (61, 66, 82, 85). As shown in Fig. 3, there are various options available for ultrasound and each has different specifications. It is therefore important to understand the capability and the limitation of the ultrasound system.

Figure 3.

Ultrasound brand distribution. Brands used in less than 15 articles were classified as Others (including Acuson, Advanced Technology Laboratories, Alpinion, Ardent Sound, Brüel & Kjær Medical, Canon, Chison, Clarius, Diasonics, Diasonograph, Dynamic Imaging, Fujifilm, Fukuda Denshi, Honda, Medison, Mindray, Pie Data Medical, Seiko, SuperSonic Imagine, Technicare, Terason, Ultrasonix, and Xuzhou Leo Medical Equipments). NR, not reported.

ULTRASOUND TECHNOLOGY DEVELOPMENTS IN SKELETAL MUSCLE IMAGING

Early ultrasound technology, static B-mode imaging, was replaced by real-time ultrasound scanners in the 1980s and 1990s. This replacement allowed investigators to freely move the transducer to generate images in any axis while the images are continuously produced automatically. These features significantly lowered the skills and training required for obtaining images compared with static scanning (105). Although these advantages expanded the use of ultrasonography, a few beneficial features from static scanners were lost. One of these benefits was reduced image artifacts. Static B-mode scanners obtain images from ultrasound beams oriented along different directions whereas real-time B-mode images are acquired from a single angle ultrasound beam that is perpendicular to the transducer. Because each pixel on the ultrasound image varies in the appearance based on the direction of the ultrasound beam, averaging echoes obtained from multiple directions helps to cancel out the random pixel patterns, reduce the artifacts, and consequently make the images look less grainy (106–108). An ultrasound imaging method called spatial compound imaging was then introduced, offering ultrasound beams at several angles, which restores the reduction in the image artifacts. In addition, development in modern transducers equipped with high-frequency pulse waves allowed an advancement in resolution compared with early real-time B-mode scanners (109–112).

Another feature lost during the transition from static to real-time scanners was the large imaging field-of-view. The biggest issue to manage with real-time ultrasound imaging for quantitative muscle assessment is the limited field-of-view. With linear-array transducers, the most frequently used type of probe for skeletal muscle assessments, the field-of-view is limited to the footprint of the transducer. This limits single image ultrasound examinations to relatively smaller muscles, like the rectus femoris. There are essentially two methods currently used to overcome this limitation. One method is to stitch the multiple images together to construct a larger image. This method was first applied in 2004 to vastus lateralis assessment by Reeves et al. (113) with a 7.5 MHz transducer. They demonstrated that the ultrasound- and MRI-derived CSA showed excellent agreement with a mean typical error of 1.7%. This is noteworthy because much better ultrasound image resolution has become available since this investigation. Another method is extended field-of-view imaging mode, also known as panoramic imaging mode. This relatively new technique uses real-time ultrasound and automatically constructs the panoramic image as an investigator scans a large structure. Ahtiainen et al. (114) first examined this technique in 2010 to assess vastus lateralis CSA before and after 21 wk of resistance exercise training. Panoramic mode ultrasound imaging resulted in 10% (P < 0.01) smaller CSA than MRI, yet the intraclass correlation coefficient (ICC) and Bland–Altman analyses showed good agreement. In addition, panoramic ultrasound imaging successfully detected the changes in CSA with resistance exercise training that was in high agreement with MRI assessment. As extended field-of-view allows less labor-intensive processing during the image analysis for large skeletal muscles, over 70% of articles reporting vastus lateralis size since 2010 used panoramic imaging mode. Both of these methods have been successfully employed since their inception and as a result, the number of publications utilizing ultrasonography for muscle assessment has dramatically increased (Fig. 1A).

Even before the extended field-of-view technology appeared in skeletal muscle assessment, three-dimensional (3 D) ultrasound imaging was introduced for human muscle quantification in 1999 (115). Constructing a 3-D image with a 2-D ultrasound scanner by tracking the motion of the transducer resulted in a 10% (range: 0.3%–19%) volume difference compared with water displacement of cadaveric human abductor digiti minimi of the hand. With more recent development in ultrasound imaging and image construction technology, in vivo 3 D ultrasonography demonstrated a mean difference of 1.1% in medial gastrocnemius volume compared with MRI (116).

Developments in ultrasound have taken place not only in imaging technology but also in image analysis. Recent studies have attempted to automatically segment the target muscle CSA. Semi- and fully automatic segmentations have been demonstrated to have high agreement with the manual tracing of the muscle borders in biceps brachii, rectus femoris, vastus lateralis, tibialis anterior, and medial gastrocnemius (117–122).

VALIDITY OF MUSCLE SIZE ASSESSMENT WITH ULTRASOUND

The technological advancements with ultrasound have brought it into the realm of the gold standard imaging techniques for skeletal muscle size determination. As a result, comparisons of muscle size examinations between MR or CT imaging and ultrasonography have been undertaken and reported in 41 articles (Table 2). The examined skeletal muscles for these comparisons include digastric, cervical multifidus, longus colli, supraspinatus, erector spinae, lumbar multifidus, pectoralis major, rectus abdominis, abdominal oblique, iliopsoas, piriformis, sartorius, quadriceps femoris, rectus femoris, vastus intermedius, vastus lateralis, vastus medialis, biceps femoris, semimembranosus, semitendinosus, quadratus femoris, lateral gastrocnemius, medial gastrocnemius, soleus, tibialis anterior, tibialis posterior, flexor digitorum longus, peroneus longus, and peroneus brevis muscles. No studies have investigated the validity of ultrasound assessment of skeletal muscles in the arms against MR or CT imaging. The most commonly examined muscle group is the quadriceps femoris (or its subcomponents) followed by the calf muscles. Age of the subjects in these studies ranged from 3 mo to 85 yr old. There were 17 different ultrasound brands used that were equipped with 2.5–18 MHz transducers. Ten articles reported the use of either a probe holder or probe support to guide the transducer during the imaging (102, 111, 114, 127, 129, 131, 138, 145, 152, 154), but did not specifically address the potential effects of using these guides. Five studies examined the muscle size before and after 14–180 days of unloading (102, 129, 138, 152, 153). In addition, four studies assessed the muscle size before and after 10–21 wk of exercise training (114, 137, 142, 153).

Table 2.

Validity of muscle CSA and volume measurements with ultrasound vs. MRI or CT

Authors	Year	Subjects n, Age (yr)	Ultrasound Brand (Mode) Frequency	Muscles	Authors’ Conclusion on Validity
Sipilä and Suominen (123)*	1993	36 W, 74 ± 5	Aloka (SC) 5 MHz	Quadriceps	Not stated, ↑ correlation, but ↑CSA with CT (r, t test)
Hides et al. (110)	1995	10 W, 21–31	Acuson (RT) 7.5 MHz	Lumbar multifidus	Valid, if adhere to strict protocol (ANOVA)
Walton et al. (124)	1997	6 M 8 W, 19–41	Technicare (SC) 5 MHz	Quadriceps	Not stated, no difference to MRI (LOA, Wilcoxon)
Juul-Kristensen et al. (125)	2000	7 W, 27–54	Diasonics (RT) 5 MHz	Supraspinatus	Valid, high, and satisfactory agreement (EIV)
Bemben (126)	2002	5 M 5 W, 26 ± 8	FD (RT) 5 MHz	Rectus femoris	Valid, yet extreme care must be taken
Esformes et al. (127)	2002	3 M 3 W, 23 ± 3	ATL (RT) 7.5 MHz	Tibialis anterior	Valid, some disadvantages over MRI (LOA, R2)
Reeves et al. (113)	2004	3 M 3 W, 77 ± 3	ATL (RT) 7.5 MHz	Vastus lateralis	Valid, no systematic bias (ICC, R2, TE)
Lee et al. (128)	2007	6 M 4 W, 28 ± 4	ATL (RT) 10 MHz	Cervical multifidus	Not valid, variation, and small muscle size (R2)
Arbeille et al. (129)	2009	32 W, 32 ± 1	Esaote (RT) 5–10 MHz	Quadriceps§	Valid, accurate evaluation of volume change (r)
Barber et al. (116)	2009	5 M 5 W, 26 ± 5	Telemed (3 D) 10 MHz	Medial gastrocnemius	Valid, errors in volume (<2%) were negligible (LOA)
Cagnie et al. (130)	2009	14 M 13 W, 22 ± 1	Esaote (RT) 12 MHz	Longus colli	Questionable, wide LOA and large variability (LOA, t test)
Fukumoto et al. (131)	2009	6 M 4 W, NR	Honda (SC) 5 MHz	Thigh muscles	Valid, CSA tends to be smaller unless corrected (r)
MacGillivray et al. (91)	2009	11, 24–85	DI (3 D) 5–10 MHz	Rectus femoris	Valid, more rapid scanning than MRI (ICC, LOA, t test)
Ahtiainen et al. (114)	2010	27 M, 26 ± 4	GE (EFV) 10 MHz	Vastus lateralis	Valid, yet it tends to underestimate CSA (ICC, LOA)
Mendis et al. (132)	2010	5 M 4 W, 24 ± 4	GE (RT) 4 MHz	Rectus femoris Iliopsoas Sartorius	Valid, strict measurement protocol (ANOVA, ICC)
Noorkoiv et al. (111)*	2010	6 M, 29 ± 5	Aloka (EFV) 10 MHz	Quadriceps	Valid, depends on the scan location (ICC, R2, SEM)
Scott et al. (102)	2012	8 M 1 W, 35 ± 8	Philips (EFV) 9 MHz	Rectus femoris Vastus lateralis Gastrocnemius	Valid, CSA and volume trend smaller (LOA, t test)
Macrae et al. (133)	2013	2 M 9 W, 20–42	Philips (RT) 5–12 MHz	Digastric	Valid, superior to MRI in some aspects (r)
Lixandrão et al. (134)	2014	21 M 10 W, 52 ± 16	Philips (RT) 7.5 MHz	Vastus lateralis	Valid, highly heterogeneous group (LOA, r, TE)
Mayes et al. (135)	2015	5 M 6 W, 30 ± 12	Toshiba (RT) 4 and 5 MHz	Quadratus femoris	Valid, yet ↑ CSA with MRI (ICC, LOA, r, t test)
Herskind et al. (136)	2016	2 M 1 W, 5–6	Telemed (RT) 10 MHz	Medial gastrocnemius	Valid, volume differed less than 2 cm3
Lixandrão et al. (137)	2016	8 M 6 W, 63 ± 4	Philips (RT) 7.5 MHz	Vastus lateralis	Valid, no difference to MRI (MM)
Scott et al. (138)	2017	26 M 1 W, 35 ± 8	Philips (EFV) 9 MHz	Quadriceps Gastrocnemius	Valid, except for gastrocnemius hypertrophy (CCC, Δscore, LOA)
Sions et al. (139)	2017	5 M 15 W, 60–85†	Esaote (RT) 2.5–7 MHz	Lumbar multifidus	Valid, 3 image average recommended (ICC, t test)
Kretić et al. (140)	2018	33 M 54 W, 62 ± 10†	Aloka (RT) 5–13 MHz	Supraspinatus	Not stated, strong correlation (r)
Barber et al. (141)	2019	11 M 7 W, 8 ± 1†	Telemed (3 D) 9 MHz	Triceps surae	Valid, children with cerebral palsy (LOA)
Gould et al. (142)	2019	14 M 22 W, 62 ± 12†	Hitachi (RT) 7.5 MHz	Rectus femoris§	Valid, + association, small bias, wide LOA (LOA, R2, rho)
Noorkoiv et al. (143)	2019	4 M 2 W, 30 ± 9 6 M 6 W, 14 ± 3†	Telemed (3 D) 10 MHz	Medial gastrocnemius	Valid, choice of clinical trials (LOA, R2, SEE, t test)
Zhang et al. (144)	2019	31 M 28 W, 49 ± 15†	GE (RT) NR	Piriformis	Not stated, ↑ CSA with MRI due to measuring plane
Franchi et al. (145)	2020	49 M 36 W, 15 ± 1	SSI (EFV) 5–18 MHz	Hamstrings	Valid, experienced rater preferred (LOA, r)
Kositsky et al. (146)	2020	6 M, 25 ± 3	Canon (EFV) 8 MHz	Hamstrings	Valid, only at sites of large CSA (LOA, r)
Nijholt et al. (147)	2020	5 M 9 W, 28–43‡	Honda (RT) 5 and 10 MHz	Rectus femoris	Valid, if whole CSA is captured (ICC, LOA, t test)
Betz et al. (148)	2021	7 M 9 W, 33 ± 11	GE (EFV) 2.5–9.5 MHz	Vastus lateralis	Valid, validity ↓ with ↑ subq adipose tissue (ICC, R2)
Geneen et al. (149)	2021	18 M 2 W, 57 ± 17†	SonoSite (RT) 7.5 MHz	Vastus lateralis	Valid, peritoneal dialysis patients (ICC, LOA, TE)
Hernández-Belmonteet al. (150)	2021	16 M, 27 ± 11	GE (EFV) 8–13 MHz	Pectoralis major	Valid, ↓ errors with trained technician (LOA, r, SEE)
Sahathevan et al. (151)*	2021	15 M 11 W, 57 ± 10†	GE (RT) NR	Rectus femoris	Valid, hemodialysis patients (ICC, LOA)
Scott et al. (152)	2021	10 M 1 W, 48 ± 6	GE (EFV) 9 MHz	Quadriceps Gastrocnemius	Valid, matches to MRI ΔCSA during spaceflight
Sponbeck et al. (112)	2021	10 M 10 W, 34 ± 17	GE (RT) 8–12 MHz	Tibialis muscles Flexor digitorum longus Peroneus muscles	Valid, strongly correlated with MRI (LOA, R2)
Stokes et al. (153)	2021	14 M, 21 ± 3	GE (RT) 12.5 MHz	Vastus lateralis	Valid, hypertrophy, and atrophy (LOA, r)
Tanaka et al. (154)	2021	30 M, 22 ± 4	GE (EFV) 8 MHz	Rectus abdominis Abdominal oblique Erector spinae	Valid, no systematic error (ANOVA, LOA, R2)
Williams et al. (155)	2021	10, 3 mo	Telemed (3 D) 8 MHz	Triceps surae	Valid, acceptable agreement in infants (LOA)

Data are means ± SD. Ultrasound column lists ultrasound brand, imaging mode (static/spatial compound, SC; real-time B-mode, RT; extended-field-of-view mode, EFV; three-dimensional, 3 D), and transducer frequency. Authors’ conclusion on validity column lists a summary of the ultrasound validity and statistical methods used for the validation.

*The results were compared with CT scan (3 articles), all other articles compared with MRI (38 articles). †Subjects with disease conditions were included. ‡The data are interquartile range. §The same region, but different muscles were compared.

ATL, Advanced Technology Laboratories; ANOVA, analysis of variance; CCC, concordance correlation coefficient; DI, Dynamic Imaging; EIV, error-in-variables scale difference; FD, Fukuda Denshi; ICC, intra-class correlation coefficient; LOA, limit of agreement; M, men; MM, mixed model; NR, not reported; r, correlation coefficient; R², coefficient of determination; rho, population correlation coefficient; SEE, standard error of estimate; SEM, standard error of measurement; SSI, SuperSonic Imagine; TE, typical error; W, women.

Authors from 36 of the 41 studies concluded that ultrasound is a valid tool for the assessment of human skeletal muscle size in young, old, healthy, and disease individuals. Although the statistical analyses for comparison are not consistent across the studies, ultrasound assessments of skeletal muscle have showed comparable absolute muscle size and magnitude of adaptations in response to long-term exercise training and unloading compared with the gold standards. The consistent findings also suggest that there may be no body region- or muscle-specific influences on the validity. Although most studies reported muscle CSA from a single scanning point, several studies scanned multiple points of the muscle and reported an averaged muscle CSA or calculated volume (102, 114, 124, 127, 129, 136, 138, 145). Unlike MR and CT imaging, there is no automated approach available with ultrasound to control the gaps between the slices. Thus, these approaches may be more challenging with ultrasound as it may increase the chance of error with manual control of the transducer. However, these multiple scanning point methods did not seem to influence the validity. Despite the overall validity, a few studies suggested ultrasound did not corroborate MRI findings. Lee et al. (128) demonstrated that ultrasound imaging was reliable in detecting the change in cervical multifidus muscle size going from the rested to contracted state. Although the differences in CSA compared with MRI were small (0.04–0.12 cm²), the regression analysis between the two methods was not significant, possibly due to the small muscle size. In addition, Cagnie et al. (130) examined longus colli CSA and showed moderate intra- and interrater reliability. The large variability in the ultrasound obtained CSA resulted in differences from MRI derived CSA. Although these data suggest small muscle size may be more problematic, the digastric muscle with a smaller CSA has been shown to be valid against MRI (133). Thus, as highlighted by Lee et al. (128), the anatomical structure of muscle rather than the size may be more important because of the difficulty to clearly identify the lateral boundary of the muscle with both ultrasound and MRI.

In addition, it is noteworthy that the extended field-of-view imaging mode tends to produce a smaller absolute muscle size than the MRI measurement (102, 114, 146, 148). These investigators suggested potential sources of the discrepancy may be related to the panoramic image processing algorithms, anatomical structure (i.e., curvature) of the muscles, measuring plane, subcutaneous adipose tissue, and compression of muscle tissue with the transducer. For those research investigations or clinical assessments where absolute muscle size is a requirement as opposed to relative change, use of the panoramic imaging mode may require further methodological adjustments (111). Nonetheless, ultrasound can provide the accurate size of various regions of human skeletal muscles. However, several of these validation studies emphasized the importance of adhering to strict imaging protocols.

OTHER METHODOLOGICAL CONSIDERATIONS

Several other methodological considerations general to skeletal muscle imaging for size determination and specific to the use of ultrasound may influence measurement reliability. Reliability of ultrasound assessment is commonly evaluated with ICC and coefficient of variation. Ultrasound reliability has been shown to be high and generally comparable to MRI (91, 102, 110, 112, 113, 116, 126, 134, 141).

General Considerations for Skeletal Muscle Imaging

In general, fluid shifts, subject positioning, measuring point, and full muscle relaxation are key considerations for whole muscle imaging and size determination (31), and this holds true for ultrasound imaging as well. Posture-related fluid shifts is an established concept with MR and CT imaging and appears to be muscle specific, with muscles of the lower leg being more influenced than muscles of the upper leg (31). Not surprising, data from ultrasound measurements also support the notion that subject positioning and rest duration influence body fluid shifts and muscle size measurements. As real-time B-mode ultrasound imaging does not require supine or prone resting position during scanning, additional care may need to be taken for posture-related fluid shifts. Thigh muscle size is influenced by position during the scan (i.e., seated, standing, or supine) (92, 156, 157), whereas neck and lower back muscle sizes do not seem to be influenced by positioning (158, 159). In addition, the effect of rest duration after a position change has been examined in vastus lateralis and rectus femoris. Vastus lateralis CSA has been shown to decrease within the first 15 min of supine rest (156, 160), whereas the rectus femoris did not change within this time frame (92, 161). The magnitude of the vastus lateralis changes measured with ultrasound were comparable to previously reported thigh muscle data obtained with CT (31, 162). In addition, it is also important to consider fluid shifts that occur with muscular activity and implement the necessary time controls. As acute exercise induces muscle swelling, general physical activity or focused exercise training before scanning can artificially alter muscle size measurements (35, 163–165).

Ultrasound-Specific Considerations for Skeletal Muscle Imaging

Manual control of the transducer critically influences the skeletal muscle size assessment with ultrasound. The freehand feature of real-time B-mode ultrasound imaging allows for flexible imaging methodologies on various regions of the body. However, it also has the potential to lose standardization across studies. In fact, there is a lack of consistency in the methods, which are discussed in detail by other investigators (43, 44, 47).

Pressure applied on the muscle during imaging also influences muscle size determination, which is a unique challenge with ultrasound compared with other imaging methods. Although most studies report that minimal pressure is applied during imaging, an objective measure of appropriate pressure is typically not presented. Jeong et al. (166) used an ultrasound transducer fixed to a spring gauge to examine the reliability with two different pressures. Even though they did not compare the muscle size between the two pressures, an excellent ICC was shown within each pressure, suggesting that a consistent pressure may improve the reliability. In addition, Stokes et al. (153) examined the effects of force on muscle thickness with a strain gauge. When 0.5 N or higher pressures were applied, the thickness of vastus lateralis decreased compared with 0.2 N, a minimum detectable force that did not visibly depress the skin. Although these authors only examined muscle thickness, the findings provide some reference values and highlight the importance for objectively tracking the pressure.

Ultrasound examination is usually performed by experienced ultrasound technicians to reduce human error. The ability of the technician to produce reliable images is critical with ultrasound as the transducer is manually controlled by the technician. Although ultrasonography is more accessible than other imaging methods, the necessary experience is unique and it requires a consistent and well-standardized approach during the imaging. For this reason, image acquisition by multiple technicians with different levels of experience within a study may reduce the reliability of the muscle size assessment. Several studies examined the ultrasound reliability of novice technicians. Good to excellent reliability has been shown for ultrasound imaging and image analysis within novice technicians (70, 167, 168) and similar reliability compared with experienced technicians has also been shown (53, 145, 169–173). Although previous experience in image tracing appears to be advantageous (145), several hours of ultrasound training, detailed instructions, and proficiency evaluations seem to offset the lack of experience (70, 152, 168). In addition, Mueller et al. (70) reported that the time to obtain a rectus femoris image was reduced by 70% within a few measurement experiences with excellent reliability retained. Several studies that examined the validity of ultrasound against MR and CT imaging used a probe holder or a probe support to aid in image acquisition standardization (102, 111, 114, 127, 129, 131, 138, 145, 152, 154). Although no studies have examined the specific influence of these tools for validity and reliability, these standardization tools are often used and may help to obtain more reliable images, especially with less-experienced technicians. Collectively, there are several factors to consider to obtain valid and reliable ultrasound assessments of muscle size. However, the training required to obtain the necessary skills is not overly demanding or a limitation to employing the ultrasound methodology.

CONCLUDING REMARKS

This review provides an overall summary of ultrasound imaging for human skeletal muscle size assessment since its inception over 50 yr ago. Ultrasound assessment of skeletal muscle size has been shown to be valid and reliable, and the use in research investigations has grown dramatically in recent years. The utility of ultrasound imaging makes it a valuable tool for one-time and repeated measurements of skeletal muscle in a wide range of populations (e.g., young, old, athlete, healthy, and diseased) and research or clinical settings.

Although ultrasound imaging has been validated in numerous muscles, more investigations are required to expand our understanding of specific methodological approaches. These include posture- and physical activity-related fluid shifts, objective evaluation of tissue compression, transducer guide design and implementation, and technician training and skill related to image acquisition and analysis. The further development of standardized methods and technician training will expand the future applications of ultrasound in investigations of skeletal muscle size.

SUPPLEMENTAL DATA

Supplemental Table S1: https://doi.org/10.6084/m9.figshare.19004915.v1.

Supplemental Table S2: https://doi.org/10.6084/m9.figshare.19003088.v1.

GRANTS

This work and our skeletal muscle imaging research have been supported by grants from the NIH (AG-020532, AG-15833, AG-00831, AG-038576, AG-18409, AG-15486) and NASA (NNJ06HF59G, EC400-NCC9-116, NNJ04HF72G).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

M.N. and T.A.T. conceived and designed research; M.N. and T.A.T. prepared figures; M.N. and T.A.T. drafted manuscript; M.N., S.T., and T.A.T. edited and revised manuscript; M.N., S.T., and T.A.T. approved final version of manuscript.