Open Access

Musculoskeletal ultrasound: a technical and historical perspective

Ronald Steven Adler
Ronald Steven Adler
   | Nov 23, 2023
Journal of Ultrasonography's Cover Image
About this article
Previous Article
Next Article
Cite
Article Category: Review Paper
Published Online: Nov 23, 2023
Page range: e172 - e187
Received: Jul 08, 2023
Accepted: Aug 21, 2023
DOI: https://doi.org/10.15557/jou.2023.0027
Keywords
© 2023 Ronald Steven Adler, published by Sciendo
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Fig. 1.

A. Compact ultrasound system with wireless chargeable transducers. A 10 MHz linear transducer is displayed. Some of the functionality normally reserved for the scanner has been incorporated into the transducer architecture, including beam forming, steering, and some image processing. B. An early laptop system utilized in a remote setting (Ghana, 2004) for guidance for therapeutic injection(11)
A. Compact ultrasound system with wireless chargeable transducers. A 10 MHz linear transducer is displayed. Some of the functionality normally reserved for the scanner has been incorporated into the transducer architecture, including beam forming, steering, and some image processing. B. An early laptop system utilized in a remote setting (Ghana, 2004) for guidance for therapeutic injection(11)

Fig. 2.

Guided interventions with precise needle localization. A. A 25-gauge hypodermic needle (arrow) is shown adjacent to the outer epineurium of the ulnar nerve (un) for purposes of hydrodissection and perineural steroid injection. B. Anesthetic (*) surrounds the medial plantar nerve (N) during performance of a nerve block. The 25-gauge hypodermic needle is well delineated on ultrasound and monitored in a real-time mode as well as the distribution of the injectate
Guided interventions with precise needle localization. A. A 25-gauge hypodermic needle (arrow) is shown adjacent to the outer epineurium of the ulnar nerve (un) for purposes of hydrodissection and perineural steroid injection. B. Anesthetic (*) surrounds the medial plantar nerve (N) during performance of a nerve block. The 25-gauge hypodermic needle is well delineated on ultrasound and monitored in a real-time mode as well as the distribution of the injectate

Fig. 3.

Ablative procedure. A. A hypoechoic nodule (*)in a patient with a symptomatic second web space interdigital neuroma. A portion of a 17 gauge cryoablation probe (arrow) is evident passing through the center of the neuroma. B and C. Progressive development of an ice ball observed in real time, for purposes of ablation of the plantar digital nerve encompassed by the neuroma. Dense posterior acoustic shadowing followed by a curvilinear specular reflector (superficial margin of the ice ball) is apparent. Ultrasound guidance allows precise targeting of the lesion
Ablative procedure. A. A hypoechoic nodule (*)in a patient with a symptomatic second web space interdigital neuroma. A portion of a 17 gauge cryoablation probe (arrow) is evident passing through the center of the neuroma. B and C. Progressive development of an ice ball observed in real time, for purposes of ablation of the plantar digital nerve encompassed by the neuroma. Dense posterior acoustic shadowing followed by a curvilinear specular reflector (superficial margin of the ice ball) is apparent. Ultrasound guidance allows precise targeting of the lesion

Fig. 4.

Bistable image of calf rhabdomyosarcoma (1976) produced with a translating 2.5 MHz transducer(24). Printed with permission of Dr. Paul Carson. The tumor is labeled as is the tibial cortex (tibia)
Bistable image of calf rhabdomyosarcoma (1976) produced with a translating 2.5 MHz transducer(24). Printed with permission of Dr. Paul Carson. The tumor is labeled as is the tibial cortex (tibia)

Fig. 5.

A. Early B-mode ultrasound scanner with first commercially available 5 MHz linear phased array transducer. Scans of the quadriceps muscles in a soccer player (1979–1980). B. Static image (1983) shows an obliquely oriented tear in the rectus femoris (arrow). Printed with permission of Dr. Bruno Fornage
A. Early B-mode ultrasound scanner with first commercially available 5 MHz linear phased array transducer. Scans of the quadriceps muscles in a soccer player (1979–1980). B. Static image (1983) shows an obliquely oriented tear in the rectus femoris (arrow). Printed with permission of Dr. Bruno Fornage

Fig. 6.

A. 5 MHz scans of patellar tendons with knee in flexion (1982) using a homemade standoff pad to place the tendon into the focal zone of the transducer. B. Grayscale image of the left symptomatic side shows thickened hypoechoic tendon with proximal intra-tendinous calcification evident (arrow) and the right normal side for comparison. Printed with permission of Dr. Bruno Fornage
A. 5 MHz scans of patellar tendons with knee in flexion (1982) using a homemade standoff pad to place the tendon into the focal zone of the transducer. B. Grayscale image of the left symptomatic side shows thickened hypoechoic tendon with proximal intra-tendinous calcification evident (arrow) and the right normal side for comparison. Printed with permission of Dr. Bruno Fornage

Fig. 7.

Rotator cuff tears. Evolution of image quality over 20 years. A. Long-axis grayscale image of a full-thickness supraspinatus tendon tear (1985), confirmed on arthrography. Arrows indicate intact tendon. The deltoid and subcutaneous fat are labeled. Printed with permission of Dr. William Middleton. B. Grayscale image of a full-thickness supraspinatus tendon tear (1995). Significant improvements in image quality are evident. Intrinsic noise (speckle) is still apparent. Arrowheads denote the articular cartilage. A – anterior, P – posterior, d – deltoid, f – fluid. C. Grayscale image of a full-thickness supraspinatus tear (arrows) (2005). Improved image quality is evident, with speckle reduction due to spatial compounding. The internal architecture of the tissue is more apparent. d – deltoid
Rotator cuff tears. Evolution of image quality over 20 years. A. Long-axis grayscale image of a full-thickness supraspinatus tendon tear (1985), confirmed on arthrography. Arrows indicate intact tendon. The deltoid and subcutaneous fat are labeled. Printed with permission of Dr. William Middleton. B. Grayscale image of a full-thickness supraspinatus tendon tear (1995). Significant improvements in image quality are evident. Intrinsic noise (speckle) is still apparent. Arrowheads denote the articular cartilage. A – anterior, P – posterior, d – deltoid, f – fluid. C. Grayscale image of a full-thickness supraspinatus tear (arrows) (2005). Improved image quality is evident, with speckle reduction due to spatial compounding. The internal architecture of the tissue is more apparent. d – deltoid

Fig. 8.

A. Biplanar real-time grayscale image of the rotator cuff derived from a 14 MHz matrix array transducer. B. 3D acquisition of a dorsal ganglion cyst using a 14 MHz matrix array transducer shows simultaneous images of the cyst in two orthogonal planes as well as volume rendering of the cyst
A. Biplanar real-time grayscale image of the rotator cuff derived from a 14 MHz matrix array transducer. B. 3D acquisition of a dorsal ganglion cyst using a 14 MHz matrix array transducer shows simultaneous images of the cyst in two orthogonal planes as well as volume rendering of the cyst

Fig. 9.

Power Doppler (B) and color Doppler (A) ultrasound images of the biceps muscle seen in cross-section, using the same color gain settings and transducer. The color gain has been increased, resulting in color noise completely filling the color Doppler image. Even though the noise is of low power, it encompasses all possible frequency shifts. In the power Doppler (PD) image, an amplitude filter has been applied, excluding low power contributions below a fixed threshold. As a result, only the vessels are displayed, superimposed on the normal grayscale appearance of the biceps muscle
Power Doppler (B) and color Doppler (A) ultrasound images of the biceps muscle seen in cross-section, using the same color gain settings and transducer. The color gain has been increased, resulting in color noise completely filling the color Doppler image. Even though the noise is of low power, it encompasses all possible frequency shifts. In the power Doppler (PD) image, an amplitude filter has been applied, excluding low power contributions below a fixed threshold. As a result, only the vessels are displayed, superimposed on the normal grayscale appearance of the biceps muscle

Fig. 10.

Synovial hyperemia on power Doppler (PD) ultrasound using the first commercially available system with PD capability in the elbow of a patient with rheumatoid arthritis (1994). A. Extensive hypoechoic soft tissue (arrows) representing inflammatory pannus is evident. B. On PD, marked hyperemia present in the hypoechoic soft tissue is compatible with active synovitis. C. One of the first demonstrations of response to therapy using power Doppler ultrasound (1995) in a patient with septic bursitis at initial presentation. D. Two weeks after surgical incision and drainage and placement on antibiotics. The same Doppler parameters were used in both images. There was a marked decrease in the extent of hyperemia, even though the grayscale appearance continued to be abnormal
Synovial hyperemia on power Doppler (PD) ultrasound using the first commercially available system with PD capability in the elbow of a patient with rheumatoid arthritis (1994). A. Extensive hypoechoic soft tissue (arrows) representing inflammatory pannus is evident. B. On PD, marked hyperemia present in the hypoechoic soft tissue is compatible with active synovitis. C. One of the first demonstrations of response to therapy using power Doppler ultrasound (1995) in a patient with septic bursitis at initial presentation. D. Two weeks after surgical incision and drainage and placement on antibiotics. The same Doppler parameters were used in both images. There was a marked decrease in the extent of hyperemia, even though the grayscale appearance continued to be abnormal

Fig. 11.

Power Doppler (PD) ultrasound image (B) and microvascular flow (Slow Flow TM) image (A) of a soft tissue sarcoma within the rectus femoris muscle using identical Doppler parameters. Significant improvement in sensitivity and vascular morphology without blooming artifact is evident in the microvascular flow image. The mass appears hypovascular on PD, whereas the Slow Flow image depicts a hypervascular mass. C. Patient with swollen left 3rd PIP joint and history of psoriasis. PD (left) and Slow Flow image (right) long-axis views of the dorsal recess. There is mild distension of the dorsal capsule by hypoechoic soft tissue. Minimal periarticular vascularity is depicted on PD. D. On Slow Flow, there is marked synovial and periarticular hyperemia
Power Doppler (PD) ultrasound image (B) and microvascular flow (Slow Flow TM) image (A) of a soft tissue sarcoma within the rectus femoris muscle using identical Doppler parameters. Significant improvement in sensitivity and vascular morphology without blooming artifact is evident in the microvascular flow image. The mass appears hypovascular on PD, whereas the Slow Flow image depicts a hypervascular mass. C. Patient with swollen left 3rd PIP joint and history of psoriasis. PD (left) and Slow Flow image (right) long-axis views of the dorsal recess. There is mild distension of the dorsal capsule by hypoechoic soft tissue. Minimal periarticular vascularity is depicted on PD. D. On Slow Flow, there is marked synovial and periarticular hyperemia

Fig. 12.

Contrast agents. A. Microbubbles made up of a gas core with a flexible, biocompatible containment shell which is usually phospholipid but can also be a protein, such as albumin. Bubble size typically varies between 2–10 microns. B. The insonating beam operates at the resonant frequency, which is dependent on bubble size. Microbubbles in the ultrasound field react to pressure by changing size, based on the pressure amplitude of the insonating beam as indicated. C. The resultant scattered echoes are non-linear, which can be represented as an expansion consisting of the fundamental along with higher harmonic components. The goal of contrast imaging is to remove the fundamental component (e.g. tissue contribution) and image the second harmonic. Images printed with permission of Siemens medical systems
Contrast agents. A. Microbubbles made up of a gas core with a flexible, biocompatible containment shell which is usually phospholipid but can also be a protein, such as albumin. Bubble size typically varies between 2–10 microns. B. The insonating beam operates at the resonant frequency, which is dependent on bubble size. Microbubbles in the ultrasound field react to pressure by changing size, based on the pressure amplitude of the insonating beam as indicated. C. The resultant scattered echoes are non-linear, which can be represented as an expansion consisting of the fundamental along with higher harmonic components. The goal of contrast imaging is to remove the fundamental component (e.g. tissue contribution) and image the second harmonic. Images printed with permission of Siemens medical systems

Fig. 13.

Time-intensity curve. Composite image obtained from a contrast study of a patient three months out from rotator cuff repair. The upper right image shows a long-axis image of the supraspinatus tendon repair with a single suture anchor evident. The upper left image displays a single frame from a contrast study with three ROIs around the suture anchor site, proximal tendon and peribursal soft tissues. A time-intensity curve (bottom) is depicted for each ROI, using the same color scheme as the ROI boundaries
Time-intensity curve. Composite image obtained from a contrast study of a patient three months out from rotator cuff repair. The upper right image shows a long-axis image of the supraspinatus tendon repair with a single suture anchor evident. The upper left image displays a single frame from a contrast study with three ROIs around the suture anchor site, proximal tendon and peribursal soft tissues. A time-intensity curve (bottom) is depicted for each ROI, using the same color scheme as the ROI boundaries

Fig. 14.

Speckle distribution in 2D image consists of sub-resolution scatterers that are randomly distributed in space. A small lateral translation by the transducer (arrow, left image) results in a new speckle field. Provided the translation is sufficiently small (within a correlation length), the speckle can be used to estimate the degree and angle of translation (right image), allowing for image registration. Alternatively, if the displacements are too large, the images are uncorrelated. In the case of spatial compounding, where a rotation is involved, the angular displacements are sufficiently large for the speckle to decorrelate, and only the specular reflectors contribute to the final image
Speckle distribution in 2D image consists of sub-resolution scatterers that are randomly distributed in space. A small lateral translation by the transducer (arrow, left image) results in a new speckle field. Provided the translation is sufficiently small (within a correlation length), the speckle can be used to estimate the degree and angle of translation (right image), allowing for image registration. Alternatively, if the displacements are too large, the images are uncorrelated. In the case of spatial compounding, where a rotation is involved, the angular displacements are sufficiently large for the speckle to decorrelate, and only the specular reflectors contribute to the final image

Fig. 15.

Spatial compounding. Both images display high-grade partial-thickness rotator cuff tears which are similar in morphology. The image labeled B is obtained with spatial compounding, while the image labeled A does not employ any spatial compounding. While both images are diagnostic, anatomic planes and tissue morphology are more distinct in the spatially compounded image. Cortical surfaces, tissue planes, and boundaries of the tear itself are better defined
Spatial compounding. Both images display high-grade partial-thickness rotator cuff tears which are similar in morphology. The image labeled B is obtained with spatial compounding, while the image labeled A does not employ any spatial compounding. While both images are diagnostic, anatomic planes and tissue morphology are more distinct in the spatially compounded image. Cortical surfaces, tissue planes, and boundaries of the tear itself are better defined

Fig. 16.

Extended field-of-view (EFOV) images of the Achilles tendon. A. The image is obtained from the first commercially available system that had EFOV capability. The full extent of the Achilles tendon is depicted, displaying mild enlargement proximal to its insertion on the calcaneus. B. The image of a torn Achilles tendon is obtained on a later generation scanner with speckle reduction. Again, the full extent of the tear and retracted fractured enthesophyte are readily depicted. The latter displays posterior acoustic shadowing. EFOV imaging allows an overall improved gestalt view of the pathology relative to conventional small field-of-view images
Extended field-of-view (EFOV) images of the Achilles tendon. A. The image is obtained from the first commercially available system that had EFOV capability. The full extent of the Achilles tendon is depicted, displaying mild enlargement proximal to its insertion on the calcaneus. B. The image of a torn Achilles tendon is obtained on a later generation scanner with speckle reduction. Again, the full extent of the tear and retracted fractured enthesophyte are readily depicted. The latter displays posterior acoustic shadowing. EFOV imaging allows an overall improved gestalt view of the pathology relative to conventional small field-of-view images

Fig. 17.

Patient with retrocalcaneal pain. Long-axis grayscale image of the Achilles tendon insertion (left) shows prominent enthesopathic mineralization. The deep surface of the tendon appears slightly hypoechoic relative to the superficial fibers, suggesting tendinosis. A rendition of the transducer is positioned over the distal tendon and low-amplitude compression/relaxation cycles are simulated to obtain an elastogram(92). Compression elastogram (right) with image obtained during the mid-relaxation phase. The strain map is green in the deep fibers of the tendon (black arrows), corresponding to the hypoechoic region seen on the conventional grayscale image. There is improved contrast on the elastogram, and the degree of tendon softening appears more extensive in comparison. The standard color map is used, wherein higher strains (softer) are denoted in red hues, and low strain (stiff) in blue and green hues
Patient with retrocalcaneal pain. Long-axis grayscale image of the Achilles tendon insertion (left) shows prominent enthesopathic mineralization. The deep surface of the tendon appears slightly hypoechoic relative to the superficial fibers, suggesting tendinosis. A rendition of the transducer is positioned over the distal tendon and low-amplitude compression/relaxation cycles are simulated to obtain an elastogram(92). Compression elastogram (right) with image obtained during the mid-relaxation phase. The strain map is green in the deep fibers of the tendon (black arrows), corresponding to the hypoechoic region seen on the conventional grayscale image. There is improved contrast on the elastogram, and the degree of tendon softening appears more extensive in comparison. The standard color map is used, wherein higher strains (softer) are denoted in red hues, and low strain (stiff) in blue and green hues

Fig. 18.

A. Representation of shear wave acquisition on normal skeletal muscle in long axis. A focused push pulse (acoustic radiation force impulse or ARFI) results in localized momentum transfer to the adjacent soft tissue. This in turn generates a cylindrically symmetric shear wave. Subsequently, the transducer elements are used to generate tracker pulses at a high frame rate. B. Generated shear wave produced through speckle tracking at two different transducer elements (point quantification) separated by approximately 2 mm. The time (T) between peaks or troughs is estimated using a correlation-based algorithm, from which the propagation speed c is estimated (c = 2/T). A parametric map of shear wave speed is generated for the entire ROI (center map) with a color scale depicting soft (blue) to stiff (red) tissues, opposite to what is typically used in compression-based schemes. A quality factor map (far right) based on signal-to-noise ratio and correlation coefficient is included. Accuracy is determined by sampling between shear waves. If the peaks are well-defined, as in this case, without broadening, the time domain correlation gives accurate estimates
A. Representation of shear wave acquisition on normal skeletal muscle in long axis. A focused push pulse (acoustic radiation force impulse or ARFI) results in localized momentum transfer to the adjacent soft tissue. This in turn generates a cylindrically symmetric shear wave. Subsequently, the transducer elements are used to generate tracker pulses at a high frame rate. B. Generated shear wave produced through speckle tracking at two different transducer elements (point quantification) separated by approximately 2 mm. The time (T) between peaks or troughs is estimated using a correlation-based algorithm, from which the propagation speed c is estimated (c = 2/T). A parametric map of shear wave speed is generated for the entire ROI (center map) with a color scale depicting soft (blue) to stiff (red) tissues, opposite to what is typically used in compression-based schemes. A quality factor map (far right) based on signal-to-noise ratio and correlation coefficient is included. Accuracy is determined by sampling between shear waves. If the peaks are well-defined, as in this case, without broadening, the time domain correlation gives accurate estimates

Fig. 19.

Shear wave elastography (SWE) in a normal volunteer (A, B) versus a patient with and underlying chronic myopathic disorder (C, D). Images obtained from the medial gastrocnemius muscle in long axis and corresponding SWE parametric image. Additional point quantifications listed from randomly placed samples within each ROI. Images from an asymptomatic 33-year-old female show shear wave elastogram (A) and normal muscle morphology on grayscale (B) with shear wave speeds > 2 m/s up to 3 m/s (not shown). Shear wave elastogram in a 39-year-old female patient with a myopathic disorder (C) and corresponding grayscale ultrasound (D). The shear wave speeds are less than 2 m/s (not shown). The corresponding grayscale image (D) depicts a diffusely echogenic muscle with loss of normal fascicular morphology, compatible with atrophy
Shear wave elastography (SWE) in a normal volunteer (A, B) versus a patient with and underlying chronic myopathic disorder (C, D). Images obtained from the medial gastrocnemius muscle in long axis and corresponding SWE parametric image. Additional point quantifications listed from randomly placed samples within each ROI. Images from an asymptomatic 33-year-old female show shear wave elastogram (A) and normal muscle morphology on grayscale (B) with shear wave speeds > 2 m/s up to 3 m/s (not shown). Shear wave elastogram in a 39-year-old female patient with a myopathic disorder (C) and corresponding grayscale ultrasound (D). The shear wave speeds are less than 2 m/s (not shown). The corresponding grayscale image (D) depicts a diffusely echogenic muscle with loss of normal fascicular morphology, compatible with atrophy

Fig. 20.

Tissue non-linearity. A simulated sinusoidal wave from the transducer is continuously distorted at increasing depths due to variation in the speed of sound in soft tissue. In the compressional phase, the propagation speed increases, whereas it decreases during decompression. As the wave propagates, it undergoes progressive distortion from a sinusoidal wave in the near field to a sawtooth pattern at increasing depth. This distorted wave can be represented as a Fourier series containing higher harmonic contributions, the dominant being the second harmonic (twice the transmit or fundamental frequency). The backscattered second harmonic is detected, allowing higher spatial resolution with less attenuation
Tissue non-linearity. A simulated sinusoidal wave from the transducer is continuously distorted at increasing depths due to variation in the speed of sound in soft tissue. In the compressional phase, the propagation speed increases, whereas it decreases during decompression. As the wave propagates, it undergoes progressive distortion from a sinusoidal wave in the near field to a sawtooth pattern at increasing depth. This distorted wave can be represented as a Fourier series containing higher harmonic contributions, the dominant being the second harmonic (twice the transmit or fundamental frequency). The backscattered second harmonic is detected, allowing higher spatial resolution with less attenuation

Fig. 21.

Tissue harmonic imaging (THI). Images of the rotator cuff with (A) and without (B) THI, using an 18L6 MHz linear transducer (Siemens Sequoia, Siemens Medical Systems). No spatial compounding is applied for purposes of comparison. An improvement in soft tissue detail is evident with tissue harmonics applied. The axial and contrast resolution are improved at the higher harmonic components. Tissue boundaries are more distinct, and the speckle spot size appears smaller
Tissue harmonic imaging (THI). Images of the rotator cuff with (A) and without (B) THI, using an 18L6 MHz linear transducer (Siemens Sequoia, Siemens Medical Systems). No spatial compounding is applied for purposes of comparison. An improvement in soft tissue detail is evident with tissue harmonics applied. The axial and contrast resolution are improved at the higher harmonic components. Tissue boundaries are more distinct, and the speckle spot size appears smaller

Fig. 22.

Planning for sacroiliac (SI) joint injection using an axial MR image of the pelvis for registration (left). An electromagnetic field generator is used with sensors placed on the transducer, sometimes on the patient and on the needle to help localize these in space. Common fiduciary markers are used to achieve registration. Usually, a minimum of three points are chosen. The transducer position, corresponding MR and US images are displayed as well as the juxtaposition of MR and ultrasound images. In the right images, a needle trajectory is simulated and superimposed on the ultrasound and composite ultrasound/MR images. The needle trajectory is seen to enter the joint proper on the juxtaposed image
Planning for sacroiliac (SI) joint injection using an axial MR image of the pelvis for registration (left). An electromagnetic field generator is used with sensors placed on the transducer, sometimes on the patient and on the needle to help localize these in space. Common fiduciary markers are used to achieve registration. Usually, a minimum of three points are chosen. The transducer position, corresponding MR and US images are displayed as well as the juxtaposition of MR and ultrasound images. In the right images, a needle trajectory is simulated and superimposed on the ultrasound and composite ultrasound/MR images. The needle trajectory is seen to enter the joint proper on the juxtaposed image

Fig. 23.

Mass segmentation. Automatic characterization of soft tissue masses requires boundary detection as a first step. Comparison of a manually segmented boundary (A), segmentation performed using an AI platform (B) derived from the hypoechoic mass seen on the bottom left (C). Segmentation was performed with five-fold cross-validation using nnU-Net framework(128)
Mass segmentation. Automatic characterization of soft tissue masses requires boundary detection as a first step. Comparison of a manually segmented boundary (A), segmentation performed using an AI platform (B) derived from the hypoechoic mass seen on the bottom left (C). Segmentation was performed with five-fold cross-validation using nnU-Net framework(128)
eISSN:
2451-070X
Language:
English
Publication timeframe:
4 times per year
Journal Subjects:
Medicine, Basic Medical Science, other
Journal RSS Feed