I-Spin live: An open-source software based on blind-source separation for real-time decoding of motor unit activity in humans

Abstract

Decoding the activity of individual neural cells during natural behaviours allows neuroscientists to study how the nervous system generates and controls movements. Contrary to other neural cells, the activity of spinal motor neurons can be determined non-invasively (or minimally invasively) from the decomposition of electromyographic (EMG) signals into motor unit discharge activities. For some interfacing and neuro-feedback investigations, EMG decomposition needs to be performed in real-time. Here, we introduce an open-source software that performs real-time decoding of spinal motor neurons using a blind-source separation approach for multichannel EMG signal processing. Separation vectors (motor unit filters) are identified for each motor unit from a baseline contraction and then re-applied in real-time during test contractions. In this way, the discharge activity of multiple motor units can be provided as visual feedback in real-time. We provide a complete framework with guidelines and examples of recordings to guide researchers who aim to study movement control at the motor neuron level. We tested the software on data collected using either grids of surface electrodes or intramuscular electrode arrays from five lower limb muscles (gastrocnemius lateralis and medialis, vastus lateralis and medialis, and tibialis anterior). We assessed how the muscle, or variation of contraction intensity between the baseline contraction and the test contraction impacted the accuracy of the real-time decomposition. This open-source interface provides a set of tools for neuroscientists to design experimental paradigms where participants can receive real-time feedback on the output of the spinal cord circuits.

The volitional control of individual motor units is constrained within low-dimensional neural manifolds by common inputs

“…Electromyographic (EMG) signals were recorded using surface grids of 64 electrodes from the GL and GM muscles during plantarflexion tasks and from the VL and VM muscles during knee extension tasks. We used a source-separation algorithm to decompose the EMG signals into motor unit discharge times either offline or in real-time using an open-source software developed and validated by our team 26 .…”

“…We used an open-source software developed and validated by our team to decompose the EMG signals in real-time 26 . First, EMG signals were recorded during a submaximal contraction, such that channels with artifacts and low signal-to-noise ratio could be removed from all further analyses.…”

“…This baseline contraction aimed to identify the separation vectors (motor unit filters). As highlighted in our previous validation work 26 , the accuracy of the online decomposition can be improved by manually editing the offline decomposition results. Therefore, we performed a manual editing step that consisted of i) removing the spikes close to the noise level, ii) recalculating the motor-unit filter and reapplying it over the signal, iii) adding new spikes recognized as motor unit firings, and iv) recalculating the motor-unit filter (Hug et al, 2021).…”

NeuroMechanics: Electrophysiological and Computational Methods to Accurately Estimate the Neural Drive to Muscles in HumansIn Vivo

“…Beside source separation methods, other approaches have been proposed to decompose the EMG signal, such as based on the detection of templates of MUAP waveforms (De Luca et al, 2006; Nawab et al, 2010). Consensus studies (Hug et al, 2021a; Gallina et al, 2022; Martinez-Valdes et al, 2023), tutorials (Del Vecchio et al, 2020; Avrillon et al, 2023b), open- source tools (Avrillon et al, 2023b; Valli et al, 2024) and commercial software packages (e.g., DEMUSE: https://demuse.feri.um.si/) for HDEMG acquisition, decomposition and/or spike train editing are available, with ongoing research towards fully automated methods for spike train identification (Clarke et al, 2020; Rossato et al, 2023).…”

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

“…As opposed to straightforward bEMG recording and filtering, the pipeline in Fig 1 requires decomposing HDEMG signals and manually editing the identified spike trains editing, while taking precautions to identify the full spectrum of discharging MUs for accurate predictions (Fig 6), as discussed. These challenges should be addressed with the rapid emergence of guidelines for manufacturers on HDEMG grid design [40], open-source tools for automatic HDEMG decomposition and spike trains edition [114], automated spike train identification approaches based on machine learning [115] and blindsource separation methods [116], and MU pool reconstruction methods [41].…”