NeuroMotion: Open-source Simulator with Neuromechanical and Deep Network Models to Generate Surface EMG signals during Voluntary Movement

Abstract

Neuromechanical studies investigate how the nervous system interacts with the musculoskeletal (MSK) system to generate volitional movements. Such studies have been supported by simulation models that provide insights into variables that cannot be measured experimentally and allow a large number of conditions to be tested before the experimental analysis. However, current simulation models of electromyography (EMG), a core physiological signal in neuromechanical analyses, are mainly limited to static contractions and cannot fully represent the dynamic modulation of EMG signals during volitional movements. Here, we overcome these limitations by presenting NeuroMotion, an open-source simulator that provides a full-spectrum synthesis of EMG signals during voluntary movements. NeuroMotion is comprised of three modules. The first module is an upper-limb MSK model with OpenSim API to estimate the muscle fibre lengths and muscle activations during movements. The second module is BioMime, a deep neural network-based EMG generator that receives nonstationary physiological parameter inputs, such as muscle fibre lengths, and efficiently outputs motor unit action potentials (MUAPs). The third module is a motor unit pool model that transforms the muscle activations into discharge timings of motor units. The discharge timings are convolved with the output of BioMime to simulate EMG signals during the movement. Here we also provide representative applications of NeuroMotion. We first show how simulated MUAP waveforms change during different levels of physiological parameter variations and different movements. We then show that the synthetic EMG signals during two-degree-of-freedom hand and wrist movements can be used to augment experimental data for regression. Ridge regressors trained on the synthetic dataset were directly used to predict joint angles from experimental data. NeuroMotion is the first full-spectrum EMG generative model to simulate human forearm electrophysiology during voluntary hand, wrist, and forearm movements. All intermediate variables are available, which allows the user to study cause-effect relationships in the complex neuromechanical system, fast iterate algorithms before collecting experimental data, and validate algorithms that estimate non-measurable parameters in experiments. We expect this full-spectrum model will complement experimental approaches and facilitate neuromechanical research.Author summaryNeuromechanical studies investigate how the nervous system and musculoskeletal system interact to generate movements. Such studies heavily rely on simulation models, which provide non-measurable variables to complement the experimental analyses. However, the simulation models of surface electromyography (EMG), the core physiological signal widely used in neuromechanical analyses, are limited to static conditions. We bridged this gap by proposing NeuroMotion, the first full-spectrum EMG simulator that can be used to generate EMG signals during voluntary movements. NeuroMotion integrates a musculoskeletal model, a neural network-based EMG generator, and an advanced motoneuron model. With representative applications of this simulator, we show that it can be used to investigate the variabilities of EMG signals during voluntary movement. We also demonstrate that the synthetic signals generated by NeuroMotion can be used to augment experimental data for regressing joint angles. We expect the functionality provided by NeuroMotion, which is provided open-source, will stimulate progress in neuromechanics.

Wearable Neural Interfaces: Real-Time Identification of Motor Neuron Discharges in Dynamic Motor Tasks

“…Importantly, the maximum median NMSE (∼ 0.2 n.u.) was similar in both simulated and recorded MUAPs, supporting the suitability of NeuroMotion [19] as a tool to model dynamic contractions for dynamic decomposition models. In fact, the simulated/detected MUAP changes were qualitatively similar to those reported by Holobar and collaborators [12,13] for the biceps brachi.…”

“…Shortening and lengthening contractions at constant force were simulated using a full-spectrum dynamic neuromechanical model (NeuroMotion, [19]). The open-source model combines a neural model [24] that simulates MNs' recruitment and force production; an AI-based volume conductor model (BioMime, [22]) that generates and/or morphs MUAPs based on dynamic physiological parameters; and an OpenSim upper limb musculoskeletal model to infer the dynamic parameters.…”

“…Recent advances in neuromuscular modelling offer new tools to address the validation problem. One notable example is NeuroMotion [19], which is a full-spectrum dynamic neuromechanical model that leverages physiological parameters from a myoelectric digital-twin [20], a muskuloskeletal model [21] and deep learning [22] to generate dynamic contractions with MN ground truth. While models like this one may provide previously unavailable insights on the modulation induced by dynamic contractions, it is still crucial to validate their outcomes experimentally.…”

“…Therefore, to develop a reliable and robust neural interface for long-term humanmachine interaction, we first characterised how muscle fibre shortening and lengthening affect electric potentials using the NeuroMotion model [19] and experimentally recorded data from the forearm and the wrist. To analyse the experimental data, we propose a novel MUAP tracking approach for dynamic contractions that takes into account spatial, temporal, and amplitude features.…”

Motoneuron-driven computational muscle modelling with motor unit resolution and subject-specific musculoskeletal anatomy

“…For example, in multi-muscle MSK models with MU-actuated volumetric representations of muscles, the model proposed in Fig 1 is also suited for the investigation of neural synergies between muscles, where MN clusters span across muscles [35], with strategies specific to the muscle groups [107]. In the field of neurophysiology, the model of MU population proposed in this study is also applicable for the computational generation of surface EMG signals during voluntary movement [108].…”