A Dual-Modal Approach Using Electromyography and Sonomyography Improves Prediction of Dynamic Ankle Movement: A Case Study

Abstract

For decades, surface electromyography (sEMG) has been a popular non-invasive bio-sensing technology for predicting human joint motion. However, cross-talk, interference from adjacent muscles, and its inability to measure deeply located muscles limit its performance in predicting joint motion. Recently, ultrasound (US) imaging has been proposed as an alternative non-invasive technology to predict joint movement due to its high signal-to-noise ratio, direct visualization of targeted tissue, and ability to access deep-seated muscles. This paper proposes a dual-modal approach that combines US imaging and sEMG for predicting volitional dynamic ankle dorsiflexion movement. Three feature sets: 1) a uni-modal set with four sEMG features, 2) a uni-modal set with four US imaging features, and 3) a dual-modal set with four dominant sEMG and US imaging features, together with measured ankle dorsiflexion angles, were used to train multiple machine learning regression models. The experimental results from a seated posture and five walking trials at different speeds, ranging from 0.50 m/s to 1.50 m/s, showed that the dual-modal set significantly reduced the prediction root mean square errors (RMSEs). Compared to the uni-modal sEMG feature set, the dual-modal set reduced RMSEs by up to 47.84% for the seated posture and up to 77.72% for the walking trials. Similarly, when compared to the US imaging feature set, the dual-modal set reduced RMSEs by up to 53.95% for the seated posture and up to 58.39% for the walking trials. The findings show that potentially the dual-modal sensing approach can be used as a superior sensing modality to predict human intent of a continuous motion and implemented for volitional control of clinical rehabilitative and assistive devices. Index Terms-B-mode ultrasound imaging, surface electromyography, machine learning regression, dynamic ankle dorsiflexion motion, human limb intent I. INTRODUCTIONWeakened function or dysfunction of the tibialis anterior (TA) muscle, which is the primary contributing muscle for

Ultrasound Echogenicity as an Indicator of Muscle Fatigue during Functional Electrical Stimulation

“…However, one limitation is that no muscle fatigue-indicating performance comparison between the use of US ERC and the use of sEMG during the same FES-induced muscle fatigue progression is presented in the current study. Inspired by the studies in [ 39 , 40 , 45 ], future work will investigate the FES-induce muscle fatigue indicators by using sole sEMG signal, sole US echogenicity signal, and the potential fusion of sEMG and US echogenicity signals.…”

Fused ultrasound and electromyography-driven neuromuscular model to improve plantarflexion moment prediction across walking speeds

“…First of all, only a few US imaging-derived parameters, MT from both LGS and SOL, measured muscle activation levels indirectly (a macrophysiological perspective). Other US-derived parameters, like FL and PA [ 21 , 25 , 27 , 35 ], of the MTU geometry model may further enrich the neuromuscular model. A potential difficulty here is the inability to capture a larger region of interest that contains entire muscle fascicles, mainly due to the small dimension of our US transducer.…”

“…Surface electromyography (sEMG) measures electrical potentials during asynchronous muscle neurons firings, and its amplitude and frequency positively relate to muscle activation levels. Therefore, sEMG-derived signals can be used in a Hill-type neuromuscular model (HNM) or to train a machine learning approach (model-free) to predict volitional joint moment [ 17 , 18 ] and angular position [ 19 – 21 ]. However, sEMG signals suffer from interference or cross-talking from the adjacent muscles, and an inability of measuring activations of deep-layer muscles [ 22 , 23 ].…”

“…Potentially sEMG signals and US imaging provide complementary information, and the combination between them may (1) mitigate any cross-talking effect from neighboring sEMG signals and (2) lower US imaging-derived features’ drift due to the accumulated errors from cyclic joint movement. Our recent studies have shown the advantages of using the dual-modal approach over uni-modal bio-signals (sEMG or US imaging) for ankle joint moment/motion prediction under isometric/dynamic dorsiflexion studies [ 21 , 25 , 27 ] and isometric plantarflexion study [ 34 , 35 ]. Similarly, Dick et al used both sEMG and US imaging to predict plantarflexion force during dynamic cycling tasks [ 32 ].…”

Editorial: Next Generation User-Adaptive Wearable Robots

“…Similarly, previously developed methods for robust and adaptive control of UE exoskeletons ( Kiguchi et al, 2005 ; Brahmi et al, 2018 ) could be explored in lower limb exoskeletons. In addition, multi-modal approaches including those that leverage ultrasound imaging for intent prediction ( Dhawan et al, 2019 ; Zhang et al, 2021 ) and muscle in the loop control of exoskeletons can increase adaptability during real-world use.…”