Machine learning to extract muscle fascicle length changes from dynamic ultrasound images in real-time

The use of artificial intelligence in musculoskeletal ultrasound: a systematic review of the literature

PURPOSE

To systematically review the use of artificial intelligence (AI) in musculoskeletal (MSK) ultrasound (US) with an emphasis on AI algorithm categories and validation strategies.

MATERIAL AND METHODS

An electronic literature search was conducted for articles published up to January 2024. Inclusion criteria were the use of AI in MSK US, involvement of humans, English language, and ethics committee approval.

RESULTS

Out of 269 identified papers, 16 studies published between 2020 and 2023 were included. The research was aimed at predicting diagnosis and/or segmentation in a total of 11 (69%) out of 16 studies. A total of 11 (69%) studies used deep learning (DL)-based algorithms, three (19%) studies employed conventional machine learning (ML)-based algorithms, and two (12%) studies employed both conventional ML- and DL-based algorithms. Six (38%) studies used cross-validation techniques with K-fold cross-validation being the most frequently employed (n = 4, 25%). Clinical validation with separate internal test datasets was reported in nine (56%) papers. No external clinical validation was reported.

CONCLUSION

AI is a topic of increasing interest in MSK US research. In future studies, attention should be paid to the use of validation strategies, particularly regarding independent clinical validation performed on external datasets.

First tarsometatarsal joint mobility in hallux valgus during gait: A synchronized ultrasound and three-dimensional motion capture analysis

PURPOSE

To quantify the vertical translation between the first metatarsal and medial cuneiform during the stance phase of gait in young individuals with and without hallux valgus.

DESIGN

This cross-sectional observational study included 34 young adults (male, n = 4; female, n = 30) who were divided into three groups according to the hallux valgus angle: control (< 20°, n = 13), mild hallux valgus (≥ 20° to < 30°, n = 12), and moderate hallux valgus (≥ 30°, n = 9). The mobility of the first tarsometatarsal joint was evaluated during the stance phase using B-mode ultrasound synchronized with a motion analysis system.

RESULTS

The medial cuneiform shifted more plantar during the early phase in mild hallux valgus and during the middle and terminal phases in moderate hallux valgus than in control. The severity of the hallux valgus was correlated with a trend toward plantar shift of the medial cuneiform. The first metatarsal was located more dorsal than the medial cuneiform; however, there was no significant variation. No significant differences in the peak ankle plantarflexion angle and moment were noted between the groups.

CONCLUSION

The hypermobility of the first tarsometatarsal joint, especially plantar displacement of the medial cuneiform in the sagittal plane, was found in young individuals with hallux valgus during the stance phase of gait, and the mobility increased with the severity of hallux valgus. Our findings suggest the significance of preventing hallux valgus deformity early in life.

Quantitative Muscle Fascicle Tractography Using Brightness-Mode Ultrasound

A muscle's architecture, defined as the geometric arrangement of its fibers with respect to its mechanical line of action, impacts its abilities to produce force and shorten or lengthen under load. Ultrasound and other noninvasive imaging methods have contributed significantly to our understanding of these structure-function relationships. The goal of this work was to develop a MATLAB toolbox for tracking and mathematically representing muscle architecture at the fascicle scale, based on brightness-mode ultrasound imaging data. The MuscleUS_Toolbox allows user-performed segmentation of a region of interest and automated modeling of local fascicle orientation; calculation of streamlines between aponeuroses of origin and insertion; and quantification of fascicle length, pennation angle, and curvature. A method is described for optimizing the fascicle orientation modeling process, and the capabilities of the toolbox for quantifying and visualizing fascicle architecture are illustrated in the human tibialis anterior muscle. The toolbox is freely available.

Improved Fascicle Length Estimates From Ultrasound Using a U-net-LSTM Framework

Brightness-mode (B-mode) ultrasound has been used to measure in vivo muscle dynamics for assistive devices. Estimation of fascicle length from B-mode images has now transitioned from time-consuming manual processes to automatic methods, but these methods fail to reach pixel-wise accuracy across extended locomotion. In this work, we aim to address this challenge by combining a U-net architecture with proven segmentation abilities with an LSTM component that takes advantage of temporal information to improve validation accuracy in the prediction of fascicle lengths. Using 64,849 ultrasound frames of the medial gastrocnemius, we semi-manually generated ground-truth for training the proposed U-net-LSTM. Compared with a traditional U-net and a CNNLSTM configuration, the validation accuracy, mean square error (MSE), and mean absolute error (MAE) of the proposed U-net-LSTM show better performance (91.4%, MSE =0.1± 0.03 mm, MAE =0.2± 0.05 mm). The proposed framework could be used for real-time, closed-loop wearable control during real-world locomotion.

In vivo imaging of skeletal muscle form and function: 50 years of insight

Skeletal muscle form and function has fascinated scientists for centuries. Our understanding of muscle function has long been driven by advancements in imaging techniques. For example, the sliding filament theory of muscle, which is now widely leveraged in biomechanics research, stemmed from observations made possible by scanning electron microscopy. Over the last 50 years, advancing in medical imaging, combined with ingenuity and creativity of biomechanists, have provide a wealth of new and important insights into in vivo human muscle function. Incorporation of in vivo imaging has also advanced computational modeling and allowed our research to have an impact in many clinical populations. While this review does not provide a comprehensive or meta-analysis of the all the in vivo muscle imaging work over the last five decades, it provides a narrative about the past, present, and future of in vivo muscle imaging.

What good is a measure of muscle length? The how and why of direct measurements of skeletal muscle motion

Over the past 50 years our understanding of the central role that muscle motion has in powering movement has accelerated significantly. Fundamental to this progress has been the development of methods for measuring the length of muscles and muscle fibers in vivo. A measurement of muscle fiber length might seem a trivial piece of information on its own. Yet when combined with knowledge of the properties of skeletal muscle it has proven a powerful tool for understanding the mechanics and energetics of locomotion and informing models of motor control. In this perspective we showcase the value of direct measurements of muscle fiber length from four different techniques: sonomicrometry, fluoromicrometry, magnetomicrometry, and ultrasound. For each method, we review its history and provide a high-level user's guide for researchers choosing tools for measuring muscle length in vivo. We highlight key insights that these measurements have provided, including the importance of passive elastic mechanisms and how skeletal muscle properties govern locomotor performance. The diversity of locomotor behaviors revealed across comparative studies has provided an important tool for discovering the rules for muscle function that span vertebrate locomotion more broadly, including in humans.

Advances in imaging for assessing the design and mechanics of skeletal muscle in vivo

Skeletal muscle is the engine that powers what is arguably the most essential and defining feature of human and animal life-locomotion. Muscles function to change length and produce force to enable movement, posture, and balance. Despite this seemingly simple role, skeletal muscle displays a variety of phenomena that still remain poorly understood. These phenomena are complex-the result of interactions between active and passive machinery, as well as mechanical, chemical and electrical processes. The emergence of imaging technologies over the past several decades has led to considerable discoveries regarding how skeletal muscles function in vivo where activation levels are submaximal, and the length and velocity of contracting muscle fibres are transient. However, our knowledge of the mechanisms of muscle behaviour during everyday human movements remains far from complete. In this review, we discuss the principal advancements in imaging technology that have led to discoveries to improve our understanding of in vivo muscle function over the past 50 years. We highlight the knowledge that has emerged from the development and application of various techniques, including ultrasound imaging, magnetic resonance imaging, and elastography to characterise muscle design and mechanical properties. We emphasize that our inability to measure the forces produced by skeletal muscles still poses a significant challenge, and that future developments to accurately and reliably measure individual muscle forces will promote newfrontiers in biomechanics, physiology, motor control, and robotics. Finally, we identify critical gaps in our knowledge and future challenges that we hope can be solved as a biomechanics community in the next 50 years.

Ankle exoskeleton torque controllers based on soleus muscle models

Powered exoskeletons are typically task-specific, but to facilitate their wider adoption they should support a variety of tasks, which requires generalizeable controller designs. In this paper, we present two potential controllers for ankle exoskeletons based on soleus fascicles and Achilles tendon models. The methods use an estimate of the adenosine triphosphate hydrolysis rate of the soleus based on fascicle velocity. Models were evaluated using muscle dynamics from the literature, which were measured with ultrasound. We compare the simulated behavior of these methods against each other and to human-in-the-loop optimized torque profiles. Both methods generated distinct profiles for walking and running with speed variations. One of the approaches was more appropriate for walking, while the other approach estimated profiles similar to the literature for both walking and running. Human-in-the-loop methods require long optimizations to set parameters per individual for each specific task, the proposed methods can produce similar profiles, work across walking and running, and be implemented with body-worn sensors without requiring torque profile parameterization and optimization for every task. Future evaluations should examine how human behavior changes due to external assistance when using these control models.

Age-Related Reliability of B-Mode Analysis for Tailored Exosuit Assistance

In the field of wearable robotics, assistance needs to be individualized for the user to maximize benefit. Information from muscle fascicles automatically recorded from brightness mode (B-mode) ultrasound has been used to design assistance profiles that are proportional to the estimated muscle force of young individuals. There is also a desire to develop similar strategies for older adults who may have age-altered physiology. This study introduces and validates a ResNet + 2x-LSTM model for extracting fascicle lengths in young and older adults. The labeling was generated in a semimanual manner for young (40,696 frames) and older adults (34,262 frames) depicting B-mode imaging of the medial gastrocnemius. First, the model was trained on young and tested on both young (R2 = 0.85, RMSE = 2.36 ± 1.51 mm, MAPE = 3.6%, aaDF = 0.48 ± 1.1 mm) and older adults (R2 = 0.53, RMSE = 4.7 ± 2.51 mm, MAPE = 5.19%, aaDF = 1.9 ± 1.39 mm). Then, the performances were trained across all ages (R2 = 0.79, RMSE = 3.95 ± 2.51 mm, MAPE = 4.5%, aaDF = 0.67 ± 1.8 mm). Although age-related muscle loss affects the error of the tracking methodology compared to the young population, the absolute percentage error for individual fascicles leads to a small variation of 3-5%, suggesting that the error may be acceptable in the generation of assistive force profiles.

Untethered muscle tracking using magnetomicrometry

Muscle tissue drives nearly all movement in the animal kingdom, providing power, mobility, and dexterity. Technologies for measuring muscle tissue motion, such as sonomicrometry, fluoromicrometry, and ultrasound, have significantly advanced our understanding of biomechanics. Yet, the field lacks the ability to monitor muscle tissue motion for animal behavior outside the lab. Towards addressing this issue, we previously introduced magnetomicrometry, a method that uses magnetic beads to wirelessly monitor muscle tissue length changes, and we validated magnetomicrometry via tightly-controlled in situ testing. In this study we validate the accuracy of magnetomicrometry against fluoromicrometry during untethered running in an in vivo turkey model. We demonstrate real-time muscle tissue length tracking of the freely-moving turkeys executing various motor activities, including ramp ascent and descent, vertical ascent and descent, and free roaming movement. Given the demonstrated capacity of magnetomicrometry to track muscle movement in untethered animals, we feel that this technique will enable new scientific explorations and an improved understanding of muscle function.

A deep learning method to predict ankle joint moment during walking at different speeds with ultrasound imaging: A framework for assistive devices control

Robotic assistive or rehabilitative devices are promising aids for people with neurological disorders as they help regain normative functions for both upper and lower limbs. However, it remains challenging to accurately estimate human intent or residual efforts non-invasively when using these robotic devices. In this article, we propose a deep learning approach that uses a brightness mode, that is, B-mode, of ultrasound (US) imaging from skeletal muscles to predict the ankle joint net plantarflexion moment while walking. The designed structure of customized deep convolutional neural networks (CNNs) guarantees the convergence and robustness of the deep learning approach. We investigated the influence of the US imaging's region of interest (ROI) on the net plantarflexion moment prediction performance. We also compared the CNN-based moment prediction performance utilizing B-mode US and sEMG spectrum imaging with the same ROI size. Experimental results from eight young participants walking on a treadmill at multiple speeds verified an improved accuracy by using the proposed US imaging + deep learning approach for net joint moment prediction. With the same CNN structure, compared to the prediction performance by using sEMG spectrum imaging, US imaging significantly reduced the normalized prediction root mean square error by 37.55% ( < .001) and increased the prediction coefficient of determination by 20.13% ( < .001). The findings show that the US imaging + deep learning approach personalizes the assessment of human joint voluntary effort, which can be incorporated with assistive or rehabilitative devices to improve clinical performance based on the assist-as-needed control strategy.

Quantitative evaluation of the vertical mobility of the first tarsometatarsal joint during stance phase of gait

We determined how the in vivo mobility of the first tarsometatarsal (TMT) joint can be quantified during gait. Twenty-five healthy participants (12 females) with no history of foot disorders were included. Non-invasive ultrasound (US) with a three-dimensional motion analysis (MA) system was used to evaluate the kinematic characteristics of first TMT joint during stance phase of gait. US probe was positioned longitudinally above the first TMT joint and adjusted to its proximal dorsal prominence. Gait analysis was conducted by the MA system starting with the activation of B-mode US video at 80 frames per second and 60-mm depth for simultaneous capture. During stance phase, the first metatarsal was translated dorsally with respect to the medial cuneiform, returning to a neutral level at toe-off in all subjects. During middle stance phase, the medial cuneiform was stable in males but displaced in the plantar direction in females and was the primary contributor to the differences in sagittal mobility observed between groups. Quantitatively measuring sagittal mobility of the first TMT joint could be useful for the early detection of foot abnormalities. The dynamic characteristics of the medial cuneiform during gait in healthy females may be associated with a high prevalence of hallux valgus.

A sound approach to improving exoskeletons and exosuits

[Figure: see text].