Table 1. Description of the main sEMG properties identified in SCI studies

Amplitude envelope

Commonly computed over a moving window using the root mean square, the average rectified signal, or the integrated rectified signal, or using low-pass filtering. The sEMG amplitude during voluntary contraction is decreased in muscles that are weakened as a result of reduced innervation. Conversely, sEMG may increase at rest after SCI in the event of spasms

Arora (2020); Baker (2017); Bjerkefors (2015a); Bjerkefors (2015b); Bolliger (2010); Boorman 1992); Cioni (1986); Corbett (2014); Cotey (2009); Cremoux (2016); deVargasFerreira (2012); Dietz 1998a); Dietz (1998b); Dietz (1999); Dietz (2002); Dietz (2004); Dietz (2009); Dorneles (2020); Ferris (2004); Field-Fote (2007); Fok (2020); Forrest (2008); Fung (1989); Gomez-Soriano (2010); Gordon (2009); Gordon (2010); Grippo (2011); Harkema (1997); Hornby (2006); Houldin (2011); Huang (2009); Hubli (2011); Hyun (2004); Jordanic (2016); Jordanic (2016); Kawashima (2005); Kawashima (2008a); Kawashima (2008b); Kim (2007); Kim (2015); Knikou (2009a); Knikou (2009b); Lam (2008); Li (2012a); Li (2012b); Lim and Sherwood (2005); Lim (2005); Little (1994); Liu (2014); Liu (2016); Lu (2019); Lunenburger (2006); Maegele (2002); Magnani (2016); Magnusson (1996); Marciello 1995); McKay (2004); McKay (2005); McKay (2011); Meyer (2020); Moss (2011); Muller (2006); Mummidisetty (2012); Niemeyer (2004); Oates (2020); Onushko (2007); Onushko (2008); Onushko 2010); Onushko (2011); Ovechkin (2013); Pepin (2003); Pollock (1978); Potten (1999); Prak (2015); Seelen (1991); Seelen (1997); Seelen (1998); Sherwood (1996); Sherwood (1997); Sherwood 2000); Skold (1998); Skold (2002); Squair (2016); Tang (1994); Thigpen (2009); Thomas (1997a); Thomas 1997b); Thomas (1998); Thomas (2014); Tibbett (2019); Uzun (2012); vanderSalm (2005); vanHedel (2005); Vastano (2019); Voerman (2009); Wallace (2012); Wierzbicka (1992); Wierzbicka 1996); Winslow (2015); Wirth (2008); Woolacott (2006); Wu (2005); Wu (2006); Wu (2009); Wu (2010); Zariffa (2012); Zijdewind (2003); Zijdewind (2014); Zoghi (2016); Zupan (1998).

Normalized amplitude

Amplitude expressed as a percentage of the value obtained at maximal voluntary contraction, at a point of interest in a time series, or using electrical stimulation. Normalization to MVC can be difficult to interpret after SCI because the maximal activation is itself affected by the injury. The M/RMS ratio has been used to differentiate UMN and LMN damage

Arora (2020); Bunday (2012); Calabro (2016); Cremoux (2016); Cremoux (2017); Dekker (2018); Hayes (2014); Dorneles (2020); Hornby (2009); Houldin (2011); Kim (2007); Kim (2015); Kim (2015); Lei (2018); Leroux (1999); Meyer (2020); Niemeyer (2004); Oates (2020); Prak (2015); Prilutsky (2011); Stahl (2015); Steldt (2004); Thigpen (2009); Thompson (2011); Tibbett (2019); Wallace (2012); Wang (2013); Wirth (2008); Wu (2005); Wu (2006); Wu (2010); Zijdewind (2003); Zijdewind (2012)

Peak amplitude

Maximal sEMG value in a time-series. Similar to the amplitude envelope, decreased values can be expected after SCI, depending on the pattern of injury

Britten (2018); Dobkin (1995); Gant (2019); Gorassini (2009); Heald (2017); Nesmeyanova (1974); Nesmeyanova (1976); Pepin (2003); Sherwood (1997); Skold (1998); Skold (2002); Thomas (1997); Wierzbicka (1996); Zijdewind (2012)

Onset/offset

Timing of sEMG activation defined when the signal amplitude is greater (onset) or lower (offset) than a pre-defined baseline threshold. Disruptions in the timing of muscle activation can be expected during complex movements (e.g. gait) after SCI, depending on the location and severity of the lesion

Aguiar (2018); Arora (2020); Bajd (1982); Bajd (1984); Baker (2017); Benz (2005); Beres-Jones (2003); Calancie (2006); Field-Fote (2007); Fok (2020); Forrest (2008); Fung (1989); Gorassini (2009); Harkema (1997); Hidler (2002); Kawashima (2005); Kawashima (2008); Kawashima (2008); Koshland (2005); Labruyère (2013); Leroux (1999); Liu (2016); Maegele (2002); Meyer (2020); Moss (2011); Mummidisetty (2012); Onushko (2010); Onushko (2011); Pepin (2003); Schmit (2002); Schmit (2003); Tang (1994); Tibbett (2019); vanderSalm (2005); Voerman (2009); Wallace (2012); Wierzbicka (1992); Wierzbicka (1996); Winslow (2009); Winslow (2015); Wu (2006); Wu (2009); Wu (2010)

Qualitative or semi-quantitative scores

Qualitative and semi-quantitative scores (e.g. present vs absent) based on the analysis of sEMG time-series. Semi-quantitative scoring criteria were established to approximate a manual muscle testing score, as follows: 0, no volitional recruitment or rate-modulation of spontaneously active motor units; 1, recruitment of 1 or 2 motor units, unsustained discharge; 2, recruitment of 1–3 motor units, sustained discharge, slow onset and/or offset; 3, recruitment of 3 motor units, but not enough to become indistinct audibly, sustained discharge, rapid onset and offset; 4, recruitment of multiple units, indistinguishable visually and audibly, rapid onset and offset, moderate amplitude at peak; and 5, recruitment of multiple units, indistinguishable visually and audibly, rapid onset and offset, high amplitude (> 0.1 mV) at peak

Alexeeva (1997); Calancie (1996); Calancie (1999); Calancie (2000); Calancie (2001); Calancie (2002); Calancie (2004); Calancie (2004); Chou (2005); Douglas (1989); Foldes (2017); Gildenberg (1985); Koshland (2005); Lewko (1995); Melzak (1954); Sherwood (1992)

Synergy analysis

Commonly obtained by using non-negative matrix factorization or principal component analysis to identify distinct patterns of activation across several muscles that can be combined to achieve desired movements. Synergies can be altered after SCI, for example reflecting patterns of muscle co-activation emerging after the injury

Baniasad (2018); Barroso (2016); Hayes (2014); Perez-Nombela (2017); Zariffa (2012)

Co-contraction index

Commonly obtained by the ratio of agonist to antagonist sEMG amplitude. Changes in supraspinal input and reorganization in the spinal circuitry after SCI can lead to altered patterns of co-activation

Boorman (1996); Cremoux (2017); Gomez-Soriano (2010); Gorassini (2009); Sherwood (1996); Stahl (2015); Tang (1994); Thomas (1998)

Coherence

An index used to indicate functional connections between the cortex and muscles or to indirectly measure shared influence in activating pairs of muscles (intermuscular coherence) or distinct motor pools within the same muscle (intramuscular coherence). After SCI, changes in intramuscular and intermuscular coherence at β or lower frequency bands (< 13 Hz) have been associated with changes in supraspinal inputs and spasms, respectively

Aguiar (2018); Barthélemy (2010); Bravo-Esteban (2014); Bravo-Esteban (2017); Cremoux (2017); Hansen (2005); Norton (2006)

Power spectral density (PSD)

PSD is the frequency domain power distribution of the sEMG. It is often computed using a Fast Fourier Transform. Characteristics of the PSD after SCI can be used to describe phenomena including clonus and fatigue.

Magnusson (1996); Onushko (2007); Steldt (2004); Wang (2013)

Median frequency (MDF)

Median value of the PSD of the sEMG signal, which can contribute to summarizing changes in the power spectrum

Niemeyer (2004); Umezu (2003); Uzun (2012).

Wavelets

Provides a time-frequency decomposition of the sEMG, which can be helpful in analyzing the frequency profiles of multi-stage movements (e.g. gait)

Meyer (2020); Mummidisetty (2012)

Motor unit pattern, number, and size

The number, size, and firing patterns of motor units recorded at the muscle surface. Although limited work exists on motor unit decomposition in sEMG after SCI, findings indicate altered firing patterns, reduced motor unit number, and increased size after SCI

Li (2012b); Li (2015); Winslow (2009); Witt (2020); Xiong (2008); Yang (1990)

Sample entropy

A measurement of the complexity of the sEMG time-series

Liu (2016)

Zero crossings

The number of zero crossings in the sEMG time-series

Corbett (2014)

Fourth order autoregressive coefficients

The coefficients of the fourth order autoregressive model of the sEMG time-series

Lu (2019)

Waveform length

The length of the sEMG time-series

Lu (2019)