MEDICINE

Time-Frequency Analysis of Motor-Evoked Potentialin Patients with Stroke vs Healthy Subjects: a Transcranial MagneticStimulation Study

Neha Singh1& Megha Saini1 & Nand Kumar2 & K. K. Deepak3 & Sneh Anand1,4

& M. V. Padma Srivastava5 &

Amit Mehndiratta1,4

Accepted: 18 July 2019# Springer Nature Switzerland AG 2019

AbstractConventional analysis of motor-evoked potential (MEP) is performed in time domain using amplitude and latency, which encapsulatesinformation relevant to the cortical excitability of the brain. The study investigated the importance of time-frequency analysis bycomparingMEPs in time-frequency domains (TFD) of healthy versus stroke survivors. Six healthy subjects and ten patients with strokewere enrolled. Single-pulse transcranial magnetic stimulation (TMS) at resting motor threshold (RMT) was given at extensor digitorumcommunis muscle cortical representation to obtain MEP. MEPs were obtained at resting motor threshold (100% RMT subjects andpatients), supra-threshold range (100–170% RMT), and different voluntary contractions (100% RMT) to subjects. Fast Fourier trans-form and continuouswavelet transform (CWT)were used for analysis. Frequency spectrum showed 98%and 66%of signal power in 0–100 Hz for subjects and patients, respectively. Top 10, top 25, and top 50 percentile power of CWT were calculated for each MEP.Frequency spectrum of top 10 and top 25 percentile power of subjects were different (p< 0.05) and dispersed to 0–500 Hz for patients;both groups having a 40-Hz peak. Total power ofMEPwas found to be low (p< 0.05) in patients as compared to subjects and top 10, top25, and top 50 percentile power showed decrease. Clinical scores—MAS and FM—were observed to be correlated to frequency andtime-frequency features (p< 0.05). Frequency spectrum belonging top 10 percentile power of different level voluntary contractionsshowed statistical significance (p< 0.05). However, no significant differences were observed for MEPs at different supra-thresholdintensities. Results suggest time-frequency analysis might provide objective ways to quantify TMS measures for stroke patients.

Keywords Stroke . Time domain features . Time-frequency domain analysis . Frequency spectrum . Extensor digitorumcommunismuscle . ModifiedAshworth Scale

Introduction

Since the discovery of magnetic stimulation of the brain, trans-cranial magnetic stimulation (TMS) has been used in numerousclinical applications with its therapeutic potential being an activeresearch area [1]. TMS has been shown to be a valuable tool instudying the regional localization [2], connectivity of brain [3],pathophysiology of neurological disorders [4], and diagnosticutility [5]. TMS technology has its advantage of being non-inva-sive, painless, good spatiotemporal resolution, and ability to con-trol the amount of stimulation intensity to be delivered [6, 7].

TMS can be used either as single-pulse TMS (sTMS), dualpulse TMS, or repetitive TMS (rTMS) protocol. Motor-evoked potential (MEP) obtained using transcranial magneticstimulation (TMS) can be recorded as electromyogram(EMG) activity in a target muscle. MEP encapsulates

This article is part of the Topical Collection on Medicine

Electronic supplementary material The online version of this article(https://doi.org/10.1007/s42399-019-00113-1) contains supplementarymaterial, which is available to authorized users.

* Amit Mehndirattaamit.mehndiratta@keble.oxon.org

1 Centre for Biomedical Engineering, Indian Institute of TechnologyDelhi (IITD), Block III, Room No: 298, Hauz Khas, NewDelhi 110016, India

2 Department of Psychiatry, All Indian Institute of Medical Sciences(AIIMS), New Delhi, India

3 Department of Physiology, All Indian Institute of Medical Sciences(AIIMS), New Delhi, India

4 Department of Biomedical Engineering, All India Institute ofMedical Sciences (AIIMS), New Delhi, India

5 Department of Neurology, All India Institute of Medical Sciences(AIIMS), New Delhi, India

https://doi.org/10.1007/s42399-019-00113-1SN Comprehensive Clinical Medicine (2019) 1:764–780

/Published online: 15 August 2019

information relevant to the cortical excitability of the brain [1]providing insights into membrane excitability of neurons,conduction and functional integrity of cortico-spinal tract,and neuromuscular junctions and is of prognostic importancein disease monitoring [2].

Two important features of MEP in time domain areamplitude and latency. Variation in time domain featuresof MEP is currently used in clinics to evaluate diseasestatus and measure treatment responsiveness [3, 4, 8].Following single-pulse TMS, the pyramidal neurons areactivated transsynaptically producing “I” indirect waves atthe threshold and as the stimulation intensity increases,pyramidal axons are activated producing “D” direct waves[5]. The level of excitability of motor cortex decides thesize of descending volleys and, hence, the amplitude ofMEP [6, 7]. Conduction time taken by neural impulses totravel along the cortico-spinal projections to peripheralmuscles is reflected in latency period of the MEP.

Quantitative measures of cortico-excitability in time do-main amplitude and latency qualify as strong input candidatesas a marker of excitability and inhibitory measures in differentapplications [1, 4]. In healthy subjects, MEP amplitude in-creases with an increase in stimulus intensity; amplitude tendsto increase almost linearly with stimulus intensity reachingplateau phase at higher stimulus intensity. The curve issigmoid-shaped in nature and referred as stimulus responsecurve (SRC) [7, 9–11].

Time domain (TD) features are sensitive to noise, arenot always accurate and might often result in false-positive and false-negatives interpretations, and have ir-reproducibility [12–15], limiting its wide acceptance inthe clinical community. It usually evidences a steeplearning curve requiring significant experience for whichvisual monitoring by trained technicians and cliniciansare needed [12]. They cannot be used in the automatedanalysis [12]. Previous studies on various evoked poten-tials in animals [12, 16–21] and humans [22] have dem-onstrated the importance of analysis in frequency domain(FD) and time-frequency domain (TFD) for differentpathophysiology conditions and showed that TFD mighthave advantages over TD analysis. Applications of FDand TFD of MEP in stroke have not been reported inhumans and animals.

In this study, we intended to perform TD, FD, and TFDa n a l y s e s o f ME P e x p l o r i n g t h e p o t e n t i a linformation which might provide improved quantitativeneurophysiological parameters. In our study for MEPanalysis, the commonly investigated forearm muscle ex-tensor digitorum communis (EDC) in stroke has beenchosen as also suggested in the literature [15, 23, 24].The functional cortico-muscular connectivity and coher-ence, till date to the best of our knowledge, has only beenstudied using EEG, MEG, and EMG [25] but not using

MEP which is a direct measurement of neuronal connec-tions and relationship between cortical and muscular ac-tivity. We also explored time-frequency information ofMEP, for healthy subjects and patients with stroke, driv-ing muscle at different threshold intensities at restingstates and different grades of isometric contraction.

Materials and Methods

Subjects and Electromyography Recording

Right-handed male healthy subjects (n = 6, age = 27.5 ±7.2 years) with inclusion criteria age 18–70 years, noneurological-deficit/hypertension/diabetes, and compliantwith TMS procedure were chosen for this study. Tenright-handed patients with stroke (n = 10, male/female =8:2, age = 49.8 ± 14.85 years, Table 1) with the followinginclusion criteria: within 24 months chronic, first, andunilateral ischemic/hemorrhagic stroke, age 18–70 yearshaving no peripheral neuropathy, Modified AshworthScale (MAS) ≤ 3, compliant with TMS procedure wereenrolled in this study. The clinical scales—MAS and up-per limb Fugl-Meyer (FM) scale—were measured for allenrolled patients (Table 1). The MAS of wrist joint wasassessed ranging from 1 to 3 on a scale of 0–4 and upperlimb FM scale ranged from 20 to 52 on a scale of 66. Thestudy was approved by the Institutional Review Board(IRB) at the All India Institute of Medical Science, NewDelhi (IEC/NP-99/13.03.2015), for healthy subjects andpatients with chronic stroke (more than 3 months to lessthan 24 months only). Written informed consent was ob-tained from all participants at the time of enrolment.

The disposable gel-based wet Ag/AgCl surface electrodeswere used in a bipolar configuration in which active electrodeswere placed on the muscle belly with a center-to-center inter-electrode distance of 20 mm and ground electrode was placedon the lateral epicondyle. Muscle contraction causing exten-sion of third digit of hand was observed for identification ofmuscle-belly and electrode placement. Electrodes were con-nected to the EMG amplifier connected with TMS (MagstimRapid2, Magstim, UK). Sampling frequency of MEP was6000 Hz, MEP was saved in .MRF (Meta Raster FormatXML metadata) format for further processing.

Experimental Setup and Procedure

Subjects were positioned comfortably on TMS recliningchair in half supine posture, with left forearm pronated,elbow joint in 90–120° flexion, wrist joint in neutral po-sition, and fingers at rest. MEP was acquired in a quietplace, the patients were instructed to close the eyes, keepthe hand in a fully relaxed condition for 120 s before

SN Compr. Clin. Med. (2019) 1: –764 780 765

starting the experiment, and take deep breaths. The experi-ment was done at the same time of the day to ensure sameexperimental conditions for all the subjects. Specific hotspotfor the EDC muscle was determined for each individual.

Single-pulse TMS were delivered to measure the MEPat cortical representation of the left EDC muscle (be-tween Cz and C4 of contralateral primary motor cortexwith reference to the EEG cap wore by them) for healthysubjects and left or right EDC muscle (between Cz andC4 or Cz and C3) for patients for the respective affectedhand [26]. TMS stimuli were delivered by a flat 70 mmfigure-of-eight coil (type D70 (AC), serial no. 0326,Magstim Rapid2, Magstim, UK), placed tangentially withhandle pointing towards back, 90° to central sulcus and45° to midsagittal line for trans-synaptical activation ofthe cortico-spinal tract [27]. TMS stimuli were deliveredby moving the coil in millimeters in all directions untilthe hotspot was localized (“hotspot” is the area produc-ing maximum MEP response for the respective muscle).Once the hotspot was localized, RMT was measured byprogressively increasing the maximum stimulator output(MSO) starting from stimulus intensity of 35% in stepsof 2 to 5% until a reliable MEP (>50 μV peak-to-peak)appears [7]. Then, MSO is lowered in steps of 1% untilthere are 5 consecutive responses out of 10 trials. Eachpulse were given at an interval of > 5 s [28]. Once theRMT was determined for a reliable MEP at the hotspot,the stability of the hotspot of EDC muscle throughoutthe experiment was ensured by marking the area with acolored marker. We encountered approximation ofhotspot in slightly lateral or posterior or lateroposterioras compared to unaffected side hotspot in the patientcohort in the study.

Healthy SubjectsMEPswere recorded by giving single-pulseTMS at the hotspot of EDCmuscle of each healthy subject in

threeways: (a) at 100%RMT, (b) at different intensity values(110%, 130%, 150%, and 170%) with increasing intensityexpressed relative to the respective RMT in supra-thresholdrange, and (c) at isometric wrist extension at 50% MVC(maximum voluntary contraction) and 100% MVC at100% RMT. Thus, in total, seven MEPs were recorded foreach healthy subject. For the second way (b), the order ofstimulus intensity was randomized for subjects avoiding an-ticipation bias inMEP recording.Also, the interval of 5 swasmaintained in between each stimulus to avoid overlaps inrefractory period of the last and action period of the nextMEP. Once the hotspot was localized and RMT was deter-mined, only the last MEP (5/10 trials) was saved (for eachseven conditions for seven subjects) due to the hardwarelimitations of Rapid2 Magstim TMS as each MEP samplefile needs to be savedmanually and individuallywhilemain-taining the coil position at the hotspot.

Patients RMTwas determined for patients on their respectiveaffected side by similar methodology as described above.MEP was recorded only at 100% RMT for patients, with sin-gle MEP for each patient.

Data Analysis

The sampling frequency of the MEP pod, built-in withTMS machine, is an effective 40-k samples per second.It is implemented using a 24-bit delta-sigma A-to-D con-verter with a 16-MHz, which gives an effective 40-k sam-ples per second at 19 bits effective dynamic range. Thesecond-order low pass anti-aliasing filter is fixed at10 kHz. The first-order high pass filter to remove DCoffset was set at 2 Hz. Electromagnetic interference(EMI) f i l ters were also implemented. And thendownsampling was done before storing the data at6 kHz. Data analysis was performed using in-house build

Table 1 Details of patients enrolled

Patients Age (years)/sex

Affected side Diagnosis Chronicity(months)

MAS scales Upper limb FM scale

Patient 1 47/M L R MCA, parietal lobe, basal ganglia ischemic 3 1+ 44

Patient 2 70/M L R MCA ischemic 3 3 20

Patient 3 61/M L R frontal ischemic 5 1+ 38

Patient 4 65/M L R pons + medulla ischemic 6 2 36

Patient 5 44/F L R thalamic + cerebellar hemorrhagic 6 1+ 31

Patient 6 53/M L R basal ganglia hemorrhagic 20 1+ 50

Patient 7 58/M R L parietal lobe ischemic 5 1+ 52

Patient 8 34/M R L gangliocapsular region ischemic 3 1+ 46

Patient 9 21/F R L temporoparietal and gangliocapsular ischemic 23 2 36

Patient 10 45/M R L basal ganglia hemorrhagic 3 2 32

R right, L left, MCA middle cerebral artery

SN Compr. Clin. Med. (2019) 1: –764 780766

algorithms in MATLAB (R2013a 8.1.0.604, Mathworks,Inc., USA). Final data sample for analysis included thefollowing: six MEP samples at 100% RMT for six healthysubjects (total six samples), four MEP samples for supra-threshold % RMT: 110%, 130%, 150%, and 170% eachfor six healthy subjects (total 24 samples), two MEP sam-ples for MVC at 50% and 100% each for six healthysubjects (total 12 samples), and ten MEP samples at100% RMT for ten patients with stroke (total ten sam-ples). Hence, total 52 MEP samples were analyzed foreach variable. The analysis window for all MEP sampleswas 49 ms that began after the TMS application to re-move the artifacts. The analysis was performed on thewhole 49 ms to remove the bias due to difference oflatency and duration in healthy, stroke, and inter-variability within stroke population.

Frequency Domain Analysis Frequency spectrum of MEP wasanalyzed using fast Fourier transform (FFT). The followingfeatures in FD were obtained: (a) power in the signal wascalculated under different bands of frequency spectrum (0–100 Hz and 0–500 Hz) and (b) full width half maximum(FWHM) of the frequency spectrum of the signal.

Time-Frequency Domain Analysis TFD analysis was per-formed using continuous wavelet transform (CWT).Grossmann and Morlet [12, 19] CWT was used as it is quiteefficient in detecting lower amplitude signals and higher spec-tral components. Minimum time-bandwidth is provided byMorlet wavelets and signal is decomposed into a set of motherwavelets on the basis of functions like contractions, shifts, anddilations. CWTwas calculated using FFT algorithm.

CWT t; að Þ ¼ ffiffiffi

ap

∫x aτð Þ*g⋅ t−τa

� �

ejtωh i

where a is scale factor for compression and expansion of themother wavelet g(t), τ denotes time shift, and * is the complexconjugate. Function g(t) represents bandpass function cen-tered around center frequency.

The following features in TFD were obtained: (a) mag-nitude of CWT coefficients; (b) total power of CWT co-efficients in signal; (c) percentile power calculations—top10 percentile, top 25 percentile, and top 50 percentile ofCWT coefficients; and (d) range of specific frequencycomponents in the signal at specific magnitudes. Highermagnitudes represented the higher contribution of the re-spective frequencies and vice versa.

Using MEP at 100% RMT, t test was used individually foreach feature of TD, FD, and TFD comparing healthy subjectsand patients for statistical significance (p < 0.05). t tests werealso used for comparing the frequency spectrum of all supra-threshold intensities in healthy subjects evaluating differences.A one-way ANOVA test was used between groups 0%, 50%,

and 100%MVC of the frequency content of top 10 percentilepower determining the statistical difference (p < 0.05).

Pearson correlation coefficients of FWHM (in FD) andfrequency content of top 10 and 25 CWT coefficients percen-tile power (in TFD) with two clinical scores (MAS and FM)were calculated individually to correlate these features withpatients’ clinical conditions.

Results

MEP was successfully recorded in six healthy subjects and 10patients enrolled in the study. Clinical scores were successful-ly measured from all these patients (see Table 1 for details ofpatients). Tables 2 and 3 show various parameters obtainedfrom time, frequency, and time-frequency domain of 6 healthysubjects and ten patients and their correlations with clinicalscores.

Time Domain Analysis

The mean MEP at 100% RMT for six healthy subjects wasobserved to be 186.44 μV peak to peak as compared to124.51 μV for ten patients (Fig. 1a, b, Table 2, AppendixTable 4). SRC with relative supra-threshold stimulus (Fig.1c) of MEP amplitude for healthy subjects showed the sig-moidal pattern of increasing MEP response with stimulus in-tensity. MEP increases sharply with stimulus intensity in 130–150% RMT, reaching a plateau with stimulus intensity >150% RMT. Supra-threshold intensities were achieved by en-hanced on option in TMS (present in Rapid2 Magstim). At50% and 100%MVC, polyphasic response was observedwithhigh amplitude (Fig. 1d).

RMT for healthy subjects (55 ± 10) showed significant dif-ferences (p = 0.015) with lower values than patients (73 ± 14).MEP amplitude for healthy subjects (186.4 ± 88 μV) was con-siderably higher than patients (124.5 ± 42.5) (p = 0.078) andlatency in healthy subjects (16.5 ± 1.1 ms) was considerablyless as compared to patients (22.5 ± 6.6 ms) (p = 0.051).

Frequency Domain Analysis

In healthy subjects, frequency range of 0–100 Hz wasobserved with clear defined peaks of 40 and 60 Hz. Inpatients, the frequency spectrum was lower in magnitudewith a spread in 0–250 Hz range, MEPs with multiplepeaks of 40, 60, 80, 160, 200, 220, and 240 Hz, buthighest peaks being at 40 and 60 Hz (Fig. 3a, b,Appendix Table 5). Table 2 compares the differencesbetween TD, FD, and TFD features of MEP at 100%RMT for healthy subjects and patients. Power analysisof frequency spectrum at 100% RMT shows significantdifferences, out of full frequency range 0–3000 Hz,

SN Compr. Clin. Med. (2019) 1: –764 780 767

approximately 82% and 66% of power of signal lies infrequency range 0–100 Hz (p = 0.09). Also, though notsignificant, 96% and 92% of power of signal lies infrequency range 0–500 Hz (p = 0.24) in healthy subjectsand patients, respectively (Table 2, Appendix Table 5).The FWHM of each MEP at 100% RMT of healthy sub-jects and patients were calculated (p = 0.971) (Table 2).Pearson correlation coefficients (CC) of FWHM wasfound to be posi t ive with MAS (CC = 0.739 atp = 0.0145) and negative with FM (CC = − 0.64 atp = 0.046) (Table 3).

Figure 3a–d shows extending of oscillations towardshigher frequency range with 0–100 Hz (peak frequency =60 Hz) at 0% MVC to 0–200 Hz (multiple clearly definedpeaks at 40 Hz, 80 Hz, and 140 Hz; peak frequency =80 Hz) at 50% MVC and 0–500 Hz (multiple clearly definedpeaks at 80 Hz, 140 Hz, 200 Hz, and 320 Hz; peak frequen-cy = 140 Hz) with a spread at higher frequencies at 100%MVC (Appendix Table 5). Figure 3e shows the frequencyspectrum of one representative healthy subject at supra-threshold stimulations. A defined pattern of frequency spec-trum with peak frequency of 40 Hz was consistently observedat all stimulus intensities from 100% RMT to 170% RMT(Fig. 3e, Appendix Table 5), and increase in the magnitudeof frequency spectrum was observed with increasing stimulusintensity similar to the SRC curve of Fig. 1c.

Time-Frequency Domain Analysis

Differences Between Healthy Subjects and Patients at 100%RMT

The presence of frequency range till 3000 Hz, shown inFig. 4 in 0–500 Hz range, with varying magnitude hasTa

ble2

Mean(±

SD)of

timedomain(TD),frequencydomain(FD),andtim

e-frequencydomain(TFD)features

RMT

(TD)

MEPam

plitu

de(TD)

Latency

(TD)

Pow

er(0–100)Hz

(FD)

Power

(0–500)Hz

(FD)

FWHM

ofsignal

(FD)

Totalp

ower

(TFD

)Pow

er(top

10perc)

(TFD)

Pow

er(top

25perc)

(TFD)

Pow

er(top

50perc)

(TFD)

Health

y55* (±

10)186.4(±

88)

16.5(±

1.1)

82.89(±

8.78)

96.5

(±4.88)

52.78(±

34.47)

1.31E+08*

(±9.8E

+07)

57.76(±

6.19)

91.51(±

3.14)

99.62(±

0.38)

Patients73* (±

14)124.5(±

42.5)

22.5(±

6.6)

66.5(±

21.05)

92.4

(±7.34)

52.26(±

52.18)

4.15E+07*

(±3.8E

+07)

55.62(±

5.57)

88.05(±

4.07)

98.93(±

1.13)

*p<0.05

Table 3 Pearson correlation coefficients (p value) of FD and TFDfeatures with clinical scores and TD features of patients

ModifiedAshworthScale

Fugl-Meyerscale

Restingmotorthreshold

MEPamplitude

Full width halfmaximum (FD)

0.73 − 0.64 0.06 − 0.58(p = 0.015)* (p = 0.046)* (p = 0.855) (p = 0.072)

Maximum contentof Frequencyrange of top 10percentilepower (TFD)

0.67 − 0.53 0.04 − 0.34(p = 0.033)* (p = 0.112) (p = 0.903) (p = 0.325)

Maximum contentof Frequencyrange of top 25percentilepower (TFD)

0.69 − 0.36 0.22 − 0.35(p = 0.025)* (p = 0.29) (p = 0.53) (p = 0.31)

*p < 0.05

SN Compr. Clin. Med. (2019) 1: –764 780768

been evidenced during the full time interval of MEP (0–50 ms) for subjects and patients (Appendix Table 6).Though different peaks with fewer magnitudes were seen inthe patients, the highest peaks were observed at 40 and 60 Hzin subjects and patients, similar to the frequency domain (Fig.3a, b). Substantial differences in subjects and patients wereobserved in terms of contributing frequency range, 0–100 Hz in healthy subjects and 0–500 Hz in patients(Fig. 4a, b). Contributing frequency range for top 10 and top25 percentile of CWT coefficients showed differences anddispersed to wider frequency range in patients as comparedto subjects (p = 0.027, p = 0.022, respectively) but not manydifferences were observed in the frequency range of top 50percentile showed (p = 0.059) (Fig. 5a, Appendix Table 6).Though the power in the contributing frequency ranges de-creased in top 10, 25, and 50 percentiles (Table 2, AppendixTable 6), total power showed significant differences amongsubjects and patients (p = 0.023) (Table 2, Appendix Table 7).

The features—maximum content of frequency rangeof top 10 and top 25 percentile—were correlated withclinical scores; correlation coefficients with spasticityscale MAS was CC = 0.672, p = 0.033 and CC = 0.696,p = 0.025, respectively (Table 3). Though positive andnegative correlations of these features were found withRMT and MEP, respectively, they did not show statisticalsignificance.

On the evaluation of frequency variations during com-plete time-duration of MEP 0–50 ms (Appendix Table 8),peak frequency of 40 Hz was prominent during peak-to-peak MEP in subjects and patients and even more insupra-threshold intensities (more than 100% RMT) insubjects (high magnitude shown in red in Fig. 4a, b andAppendix Table 8).

Time-Frequency Analysis at Different MVC

The frequency spectrum was found to be much more dispersedfor variation in 50% and 100% MVC as compared to 100%RMT at 0% MVC with frequency covering the range up toapproximately 363 Hz (Fig. 4c, Appendix Table 9). A one-wayANOVAwas applied and frequency range of top 10 percentilewas found to be different (Fig. 5b). F value for the ANOVA testwas 2784.16 with p= 1.1102E−16. All post hoc pairwise TukeyHSD comparisons suggest that the p value corresponding to theF statistic of one-wayANOVA is lower than 0.01which stronglysuggests that the frequencies are significantly different.Frequency spectrum were observed to be dispersed towards theright as % MVC is increased with clear defined multi-ple peaks shifting towards right too at 40 Hz, 80 Hz,and 140 Hz and peak frequency of 80 Hz and 140 Hzat 50% MVC and 100% MVC, respectively (Fig. 5b,Appendix Table 9).

Discussion

The conventional interpretation of TD features is potentiallydistorted by noise and can result in false-positive and false-negative results given its limitation [12–15]. The main targetof this study was to determine if there are differences in spec-tra of stroke patients as compared to that of healthy subjects inTFD. Key results of this study are as follows: with 100%RMT and the healthy subjects present an energy distributionin MEP at a certain frequency range (0–100 Hz) in TFD. Thisenergy in patients, with lowmagnitude, spreads over the range0–500 Hz and the shift in location (latency) was also observed.Mean power in frequency range (0–100 Hz) and (0–500 Hz)

Fig. 1 a Mean MEP response curve with 95% confidence interval (CI)for six healthy subjects at 100% RMT. bMeanMEP response curve with95%CI for ten patients at 100%RMT. c Stimulus Response Curve (mean

± 95%CI) of six healthy subjects with the supra-threshold stimulus (atdifferent %RMT) andMEP amplitude, showing the sigmoid shape curve.d MEP response curve at 50% MVC at 100% RMT

SN Compr. Clin. Med. (2019) 1: –764 780 769

was observed to be low in patients. The total power of CWTcoefficients and frequency range contributing to top 10and top 25 percentiles power of MEP signal showedstatistically significant differences (p < 0.05) in patients.Mean power at different percentiles (10, 25, and 50) ofCWT magnitudes were also observed to be low in pa-tients (Table 2). The FD feature -FWHM and TFDfeature- maximum frequency content in 10 and 25 per-centile of CWT magnitudes were found to be statistical-ly correlated with clinical scores of stroke patients(Table 3). The frequency range 20–70 Hz appeared inMEP might belong to piper rhythm in motor cortex.

Differences in Healthy Subjects and Patients in TimeDomain

RMT was observed to be ~ 33% higher in patients(p = 0.015) as compared to healthy subjects. MEP

amplitude was observed to be lower by ~ 50% and la-tency was higher by ~ 26% (Table 2) in patients, mightbe indicating loss of upper motor neurons (UMN) oraffected cortico-spinal tract or peripheral nerves in stroke[2, 8, 29]. The trend of SRC (Fig. 1c) as has beenobserved for different muscles was also observed in fre-quency domain analysis (Fig. 3e); amplitude was foundto increase with increase in stimulus intensity producinga higher MEP possibly because of accessing larger neu-ronal recruitment of cortico-spinal higher threshold mo-tor neurons in form of D wave [10]. SRC reaches aplateau phase at ~ 150–170% RMT (Fig. 1c) as reportedin the literature [7, 9–11], possibly being the maximumstimulus intensity for cortico-spinal neuronal firing andhighest achievable neural recruitment possible [30].None of the time domain parameters showed a correla-tion with clinical scores.

Healthy Patients

60

80

100

120

RM

T

Resting Motor Threshold

Healthy Patients

100

200

300

Ampl

itude

(uVo

lt)

MEP Amplitudes

Healthy Patients

10

20

30

Tim

e (m

sec)

MEP Latency

a

b

c

*

Fig. 2 Box and whisker plot of aresting motor threshold, b MEPamplitude, and cMEP latency forthe healthy subjects and patients;RMT shows significantdifferences in healthy subjectsand patients, but latency andMEPamplitude does not

SN Compr. Clin. Med. (2019) 1: –764 780770

Differences in Healthy Subjects and Patientsin Frequency Domain

A considerable decrease in mean power in the range of0–100 Hz was observed to be ~ 25% (82.89 to 66.5)and ~ 5% (96.5 to 92.4) in 0–500 Hz in patients ascompared to the subjects. Results reflect that the powercontent present in 0–100 Hz in subjects is considerablydispersed over 0–500 Hz in patients (Table 2, Figs.

3a, b and 4a, b). The decrease in energy content anddispersion of power towards higher frequency rangewith pathology seen here is in line with TFD studieson evoked potential conducted on rat animal model[12, 19]. The FWHM was found to be positively corre-la ted (with p = 0.01) with MAS (CC = 0.74)which might indicate towards the relationship betweenhigher spasticity with large contraction of muscle fibersand more pathology with large dispersion (right sided)

Fig. 4 Time-frequency domain analysis showing three-dimensional-viewof magnitude of coefficients of CWT at100% RMT for representative ahealthy subject (resting, 0% MVC), b patient (resting, 0% MVC), c %

RMT for representative healthy subject at 100% MVC, and d healthysubject at 170% supra-threshold intensity

Fig. 3 Mean frequency spectrum (with 95%CI) of the MEP for sixhealthy subjects at a 0% MVC (resting state, 100% RMT). b Meanfrequency spectrum (with 95%CI) of MEP curve for 10 patients (resting

state, 100% RMT). c 50% MVC, d 100% MVC, and e supra-thresholdintensities for one representative healthy subject

SN Compr. Clin. Med. (2019) 1: –764 780 771

in frequency spectrum resulting in higher FWHM andvice versa, as confirmed in the animal model [16, 19].The negative correlation (with p = 0.46) of upper limbFM scale with FWHM (CC = − 0.64) might reflect thedecrease in the dispersion of frequency spectrum withthe higher functionality of upper limb.

Differences in Healthy Subjects and Patientsin the Time-Frequency Domain

TFD analysis shows distinct information about the spec-trum of healthy subjects and patients. The features whichshowed statistically significant differences (p < 0.05)among healthy subjects and patients (Fig. 5a, Table 3)were (1) frequency spectrum of top 10 percentiles ofCWT coefficients, (2) frequency spectrum of top 25 per-centiles of CWT coefficients, (3) maximum frequencycontent of top 10 percentile of CWT coefficients, (4)

maximum frequency content of top 25 percentiles ofCWT coefficients, and (5) total power in MEP signal.These features in TFD qualify as the input candidatesfor differences between healthy subjects and patients.These findings showed that the changes in frequencyrange, in stroke patients as compared to healthy subjects,were statistically significant than the change in amplitudeand latency in time domain showing it might be a moresensitive monitoring technique. The dispersion wasshown in the frequency spectrum of top 50 percentile infull 0–3000 Hz range. A similar tendency of dispersionhas been reported in rat models evaluating variousevoked potential like MEP in spinal cord injury [16]. Inpatients, top 10, 25, and 50 percentiles power did show aconsiderable decrease (Table 2, Appendix Tables 6 and 7)and total power showed a statistically significant decreasein power by 216% (p = 0.023). The decrease might be

Fig. 5 Box and whisker plot showing the comparison of frequencyspectrum contribution at a top 10, 25, and 50 percentile (shown as prc)power of at 100% RMT frequency range between subjects and patients

and b top 10 percentile at different grades ofMVC at stimulus intensity of100% RMT, found to be different (p < 0.05 represented by an asterisk)

SN Compr. Clin. Med. (2019) 1: –764 780772

attributed to the weak cortico-spinal tract or the reducedability or loss of UMN.

The two features—maximum frequency content of atop 10 percentile of CWT coefficients and maximum fre-quency conten t of top 25 percen t i l es of CWTcoefficients—of each patient were correlated with eachpatient’s spasticity scale and showed positive correlationwith MAS. The tendency of frequency dispersion towardshigher (right-side) frequency was in confirmation of thefindings in rat animal model [16]. In humans, it could beexplained by: the more is the pathology and the more isthe spasticity, and the more is the contraction of musclefibers, having a large working frequency, controlled bylarge motor neurons and, hence, the more is the dispersionof frequency [31, 32]. Thus, stroke leads to changes intime-frequency domain and can be a better indication ofthe pathophysiology of stroke than TD as monitoring inTFD offers the detection of time shifting, frequencyshifting, frequency dispersion, power loss in contrast toamplitude, and latency in time domain.

The piper rhythm (20–70 Hz with 40 Hz peak), foundin EMG and also in frequency coherence of EMG withMEG and EEG, in differing grades of voluntary contrac-tions of muscles in healthy subjects has been reported inseveral studies [17, 25]. It has been shown to be drivenby the contralateral motor cortex and plays a critical rolein the pathophysiology of diseases [33]. The same fre-quency range with 40 Hz peak was also observed in ourstudy in healthy subjects and patients with stroke (Fig.3a, b, Appendix Table 5, 8). The range is more dispersedtowards higher frequency with lower in magnitude inpatients as compared to the healthy subjects. The pres-ence of 40 Hz rhythm in upper limb forearm musclemight indicate the presence of piper rhythm in motor-evoked potential of stroke patients, the same range hasbeen earlier reported in EMG signal from lower limbmuscle in stroke patients and in other neurological dis-ease like Parkinson disease [33, 34]. Fang et al. in 2010showed low gamma band cortico-muscular coherence inpoorly recovered stroke survivors with the help of EEG-EMG coherence [35]. This difference might be becauseof the methodological differences in the study (MEP andEEG-EMG coherence). As TMS stimulus is applied, Dwaves and I waves are recruited which sums up at ante-rior horn of the spinal cord and travels to the muscle forthe motor fiber to fire [1, 6]. So, it is worth noting thatMEP is more complex-signal that captures informationon neuronal firing, cortico-spinal tract conduction, andmuscle excitability that could also be reflected in therange and shift of frequency spectrum in our study.

Also, these differences might be because of differentchronicity, impairment, and lesion size and location.

As mentioned above, MEP is a combination of infor-mation initiating at the pyramidal axons (D waves), var-ious synapses though neuronal circuits (I waves), con-duction through the tract and muscle recruitment, limit-ing of higher frequency content (top 10 percentile ofCWT coefficients) to ~ 100 Hz in healthy subjects anddispersed to ~ 500 Hz in patients highlights substantialfundamental alterations among healthy subjects andstroke patients. One example of difference might be theunorganized recruitment of higher motor fiber firing athigher frequencies even in the resting muscles. Eventhough most of the frequencies were present at the com-plete duration of MEP, the highest CWT magnitudes arecontained by 40 Hz peak frequency lying in the periodof peak-to-peak MEP (Appendix Table 8) representingmaximum change during the peripheral evoke response.The frequency spectrum was observed to shift towardsthe right in time axis in patients (Fig. 4b) representinglonger latency period and slower cortico-spinal tract con-duction. The sample size is limited to interpret signifi-cant impact; a bigger study with similar analysis ap-proach would be highly beneficial.

The frequency range captured by the electrodes in ourstudy could contain the information of not just the musclefiring but the neuronal firing too, MEP is a complex signal,as the presence of frequency range up to 100 Hz in thebrain has also been confirmed by electrocortiocography(ECoG) study, MEG, and other in vivo pyramidal cellsstudies [6, 17, 36]. This is the first study on humans toreport 100 Hz frequency content in brain non-invasivelyusing TMS.

Hence, stroke leads to changes in frequency and time-frequency domain features—FWHM, total power, maximumfrequency content of top 10 and 25 percentile CWTmagnitudes—and can be a better indication of pathophysiol-ogy of stroke than TD as monitoring in TFD offers the detec-tion of time shifting, frequency shifting, frequency dispersion,power loss in contrast to amplitude, and latency in time do-main. The application of time-frequency domain analysismight provide an alternative way to detect/diagnose and inter-pret results in automated MEP monitoring. This approachmight be helpful for neurophysiologists by providing a prac-tical improved method to quantify TMS measures by provid-ing an advantage over traditional time domain analysis whichcould be worked out for real-time clinical applications in fu-ture research. This exploratory study opens a new pathway forautomated monitoring of neurophysiological parameters ofMEP in stroke.

SN Compr. Clin. Med. (2019) 1: –764 780 773

Differences at Different Supra-threshold Intensitiesand Different MVC in Healthy Subjects

The increase in the magnitude of oscillations (asindicated in Fig. 1c by SRC curve, shown in Fig. 3e,Appendix Table 8) with increasing stimulus intensity(signal growing vertically) might indicate a relationshipof particular frequency range with D waves being gen-erated with TMS stimulus because D waves manifeststhe same trend of increasing magnitude with increasingstimulus intensity. The magnitude increases with fre-quency range with peak frequency being constant (Fig.3e) indicating the higher cortical response due to in-crease in stimulus intensity but the same muscle recruit-ment (rate coding). This might represent the activationof higher threshold central and peripheral pyramidalneurons having faster propagations and producing largeraction potentials which finally govern large motor unitsin the target muscles [1].

The oscillation in Fig. 3a–d extends, not shifts, to-wards higher frequency range with increasing peak fre-quency at 40 Hz, 80 Hz, and 120 Hz, followingHenneman size principle [32] and indicating of firingof both slow and fast muscle fibers, with increasingMVC—0%, 50%, and 100%, respectively (Fig. 3a–d,Appendix Table 5, Figs. 4c and 5b, AppendixTable 9). The signal grows horizontally with increasingMVC and frequency range have not shifted but extend-ed showing the involvement of slow muscle fibers tooalong with fast muscle fibers in line with size principleand rate coding. The differences in the frequency con-tent of the top 10 percentile power of CWT at differentMVC was found to be statistically significant at withp = 1.1102E−16, F value = 2784 (Fig. 5b). This increasein frequency range might be due to higher neuronalfiring, as demand on the motor cortex is likely to begreater under these circumstances. Coherence analysishas revealed that some of the oscillations are transmitted,probably via the pyramidal tract, to the active muscles andmay entrain them into the same rhythmicity. The spectrumwas also observed to shift left in healthy subjects showingless latency (Fig. 4c) as MVC increases supporting thefact that motor neurons near threshold tend to dischargeexcitatory post-synaptic potential (EPSP) which summatesgiving short latency EMG [1]. For the different level ofcontractions, the cortico-muscular coherence is found tobe in piper rhythm in literature.

Limitations One of the limitations of this study is the lowsample size. It is worth noting that the acquisition of MEPis challenging in patients with stroke and this study is

exploratory in nature evaluating the potential advantagesof time-frequency analysis in MEP. Also, the age group ofthe healthy subject and patients with stroke was differentwhich can be further explored in future studies as ageaffects the cortical circuits too. Patient population had alarge amount of heterogeneity in terms of lesion location,chronicity, or even the effects of rehabilitation patientsmight be receiving since the episode of stroke. A largerstudy would get benefit with a time-frequency domainanalysis methodology proposed in the study.

Conclusion

The study demonstrates the features of the time-frequency domain of MEP analysis for healthy subjectsand patients that might be of clinical relevance in dis-ease diagnosis or prognostic monitoring. It also indi-cates the presence of piper rhythm in healthy subjectsand patients with stroke and changes in MEP frequencyrange during different MVC using non-invasive MEPwith TMS stimulation.

Acknowledgments The authors would like to express sincere gratitude tohealthy subjects and patients who agreed to participate in the study. Also,they thank Mr. Vikas Kumar and Ms. Komal at TMS laboratory for theirsupport during data acquisition and Mr. Dixit Sharma for the help in dataanalysis.

Authors’ Contribution Conceptualization: NS, AM; data curation:NS,MS; formal analysis: NS, AM; funding acquisition: AM; methodol-ogy: NS, AM; resources: NK, SA, PS; supervision: AM; writing theoriginal draft: NS, AM; writing the review and editing: NS, AM, KKD,NK, PS.

Funding This work was supported by Science and Engineering ResearchBoard (SERB), DST, Government of India (YSS/2015/000697). NehaSingh was supported with research fellowship funds from the Ministryof Human Resource and Development (MHRD), Government of India.

Compliance with Ethical Standards

Conflict of Interest The authors declare that they have no competinginterests.

Ethics Approval All procedures performed in studies involving humanparticipants were in accordance with the ethical standards of the institu-tional and/or national research committee and with the 1964 Helsinkideclaration and its later amendments or comparable ethical standards.The study was approved by the Institutional Review Board (IRB) at theAll India Institute of Medical Science, New Delhi (IEC/NP-99/13.03.2015).

Informed Consent Informed consent was obtained from all individualparticipants included in the study.

SN Compr. Clin. Med. (2019) 1: –764 780774

Table 4 Amplitude and latency of MEP response curve (100% RMT) of six subjects and ten patients

RMT MEP peak-to-peakamplitude (μV)

MEP latency (ms)

SubjectsSubject 1 43 146.1 16.1Subject 2 63 320.5 16.3Subject 3 65 84.7 15Subject 4 43 219.6 18.1Subject 5 57 234.8 16Subject 6 60 112.8 17.5Mean ± SD 55 ± 10 186.4 ± 88.0 16.5 ± 1.1

PatientsPatient 1 55 150 27.5Patient 2 65 49.7 21.6Patient 3 57 141.3 21Patient 4 62 112 6.1Patient 5 64 115 24Patient 6 93 94.1 29Patient 7 85 211 22.8Patient 8 82 150 19.1Patient 9 90 104 25Patient 10 80 117 29Mean ± SD 73 ± 14 124.5 ± 42.5 22.5 ± 6.6

Appendix

Table 5 Details of bandpower of MEP at different supra-threshold stimulus intensities (100 to 170%) RMT of subjects and patients at differentfrequency range

Stimulus intensity (% RMT) 0–100 Hz 0–500 Hz Highest peak (Hz)

Subject 1 100 93.81 97.23 40

Subject 2 100 77.92 98.77 60

Subject 3 100 76.41 86.7 40

Subject 4 100 90.486 99.049 60

Subject 5 100 87.03 99.05 20

Subject 6 100 71.71 98.64 40

Each subjects in detail

Subject 1 100 93.81 97.23 40

110 92.01 96.42 40

130 91.77 97.25 40

150 98.15 99.84 40

170 94.81 99.65 40

50% MVC 100 60 99.37 80

100% MVC 100 37.2 99.27 120

Subject 2 100 77.92 98.77 60

110 87.31 97.56 60

130 36.77 98.19 40

150 79.58 99.45 80

170 75.98 98.91 60

50% MVC 100 73 99.58 80

100% MVC 100 53.4 99.22 60

Subject 3 100 76.41 86.7 40

110 91.64 98.99 40

SN Compr. Clin. Med. (2019) 1: –764 780 775

Table 5 (continued)

Stimulus intensity (% RMT) 0–100 Hz 0–500 Hz Highest peak (Hz)

123 96.43 99.44 40

130 94.89 99.58 40

150 92.8 99.8 40

170 94.96 99.59 40

50% MVC 100 39.3 99.43 140

100% MVC 100 7.3 98.62 140

Subject 4 100 90.486 99.049 60

110 89.82 97.493 40

150 49.84 98.742 60

170 58.842 99.661 80

50% MVC 100 54.3 98.66 80

100% MVC 100 23 99.07 200

Subject 5 100 87.03 99.05 20

110 54.27 99.01 100

130 57.06 98.08 80

150 79.93 99.61 60

170 32.8 98.48 100

50% MVC 100 80 97.5 80

100% MVC 100 50.5 9723 80

Subject 6 100 71.71 98.64 40

110 72.7 97.95 60

130 73.04 99.18 40

150 49.64 98.44 40

170 83.35 99.51 80

50% MVC 100 80 98.66 80

100% MVC 100 50.5 99.07 140

Patients

Patient 1 100 87.4 97.04 60

Patient 2 100 24.5 78.81 80

Patient 3 100 71.9 98.63 60

Patient 4 100 61.05 90.48 40

Patient 5 100 73.35 92.83 60

Patient 6 100 40.8 80.43 60

Patient 7 100 65.08 97.61 40

Patient 8 100 81.98 98.77 40

Patient 9 100 94.39 97.42 60

Patient 10 100 64.9 92.34 100

SN Compr. Clin. Med. (2019) 1: –764 780776

Table 6 Details of power and frequency range of MEP at 100% RMT of subjects and patients at top 10, top 25, and top 50 percentile

Healthy subject Power (%) Frequency range (Hz) Patient Power (%) Frequency range (Hz)

H1 Top 10 percentile 58.35 (19.75–79.00) P1 62.21 (39.5–104.2)

H2 50.38 (19.75–104.26) P2 52.35 (39.5–363)

H3 51.40 (19.75–104.26) P3 49.49 (19.75–137.55)

H4 64.60 (39.50–68.77) P4 49.71 (19.75–208.49)

H5 64.75 (39.50–119.75) P5 58.49 (19.75–137.55)

H6 57.08 (39.50–208.49) P6 53.9 (19.75–119.74)

P7 58.9 (39.5–208.4)

P8 63.8 (34.38–208.4)

P9 59.05 (39.4–119.74)

P10 48.39 (39.5–416.98)

Mean (Std. Dev.) 57.76 (6.19) 55.62 (5.57)

H1 Top 25 percentile 91.88 (17.19–104.24) P1 93.24 (34.38–275.1)

H2 88.46 (17.19–239.49) P2 80.8 (19.75–416.98)

H3 88.70 (17.19–119.75) P3 88.9 (17.19–181.5)

H4 92.91 (19.75–119.75) P4 86.41 (17.19–416.98)

H5 96.86 (34.38–316.01) P5 89.7 (17.19–208.49)

H6 90.29 (19.75–275.11) P6 88 (17.19–239.49)

P7 90.04 (19.75–316.01)

P8 90.8 (19.75–363)

P9 91.17 (19.75–363)

P10 81.5 (19.75–550.21)

Mean (Std. Dev.) 91.51 (3.14) 88.05 (4.07)

H1 Top 50 percentile 99.67 (14.97–2904.04) P1 99.4 (17.19–2528)

H2 99.86 (14.97–2528.11) P2 96.5 (17.19–2904)

H3 98.85 (17.19–2904.04) P3 99.8 (14.96–2904)

H4 99.80 (17.19–416.98) P4 99.55 (14.96–2528)

H5 99.73 (17.19–2528.11) P5 99.7 (17.19–2904)

H6 99.78 (14.97–1915.95) P6 99.14 (17.19–2904)

P7 99.32 (17.19–2904)

P8 99.37 (17.19–2904)

P9 99.41 (17.19–2528)

P10 97.17 (17.19–2904)

Mean (Std. Dev.) 99.62 (0.38) 98.93 (1.13)

SN Compr. Clin. Med. (2019) 1: –764 780 777

Table 7 Details of total power ofCWT coefficients (p < 0.05) ofMEP at 100% RMT of subjectsand patients

Healthy Patients

H1 1.1E+08 P1 5.8E+07

H2 2.6E+08 P2 1.0E+06

H3 1.5E+07 P3 4.5E+07

H4 2.3E+08 P4 1.1E+08

H5 1.3E+08 P5 1.0E+08

H6 4.2E+07 P6 2.0E+07

P7 1.6E+07

P8 2.6E+07

P9 3.3E+07

P10 5.5E+06

Mean (Std. Dev) 1.31E+08 (9.8E+07) Mean (Std. Dev) 4.15E+07 (3.8E+07)

Table 8 Magnitude of 40 Hz peak frequency in MEP in time-frequency domain

Time period (0–50 ms)

Representative healthy (100%RMT)

Representative patient (100%RMT)

Representative healthy (150%RMT)

In early period 40 Hz peakfrequency

585.7 254.4 779

At latency 591.7 259.6 790.5

At positive peak 597.7 260.1 795

Baseline 599.2 260.2 798.2

At negative peak 601 260.3 799

At end of peak 602.2 260.5 791.6

Later period 587 258.8 779

Table 9 Details of frequency range at top 10 percentile ofMEP and peak frequency ofMEP response at 100%RMTand 0%, 50%, and 100%MVC forhealthy subjects

Healthy Frequency range (Hz)(0% MVC)

Peakfrequency(Hz)

Frequency range (Hz)(50% MVC)

Highest peakfrequency (Hz)

Frequency range (Hz)(100% MVC)

Highest peakfrequency (Hz)

H1 Top 10percen-tile

(19.75–79.00) 80 (59.8–158) 80 (39–239.4) 120

H2 (19.75–104.26) 60 (39.5–137.5) 80 (39.5–181.5) 60

H3 (19.75–104.26) 40 (39.5–208.4) 140 (79–363) 140

H4 (39.50–68.77) 60 (39.5–181.5) 80 (68.7–275) 200

H5 (39.50–119.75) 80 (39.5–104.2) 80 (19.75–363) 80

H6 (39.50–08.49) 40 (59.8–158.3) 80 (79–363.5) 140

SN Compr. Clin. Med. (2019) 1: –764 780778

References

1. Lefaucheur J-P, Andre-Obadia N, Antal A, Ayache SS, Baeken C,Benninger DH, et al. Evidence-based guidelines on the therapeuticuse of repetitive transcranial magnetic stimulation (rTMS). ClinNeurophysiol. 2014;125:2150–206.

2. Sliwinska MW, Vitello S, Devlin JT. Transcranial magnetic stimu-lation for investigating causal brain-behavioral relationships andtheir time course. JoVE (Journal of Visualized Experiments).2014 Jul 18(89):e51735.

3. Cantarero G, Celnik P. 11 applications of TMS to study brain con-nectivity. Brain Stimul Methodol. 2015; https://books.google.co.in/books?hl=en&lr=&id=lJc3BwAAQBAJ&oi=fnd&pg=PA191&dq=Applicat ions+of+TMS+to+Study+Brain+Connectivity.+Brain+Stimulation:+Methodologies+and+I n t e r v e n t i o n s&o t s=Y50YdyLPcR&s i g=HmJ i kFd_vZ3wtZXpOWDKEBRbBVQ. Accessed 2 Oct 2016.

4. Rossi S, Hallett M, Rossini PM, Pascual-Leone A, Avanzini G,Bestmann S, et al. Safety, ethical considerations, and applicationguidelines for the use of transcranial magnetic stimulation in clini-cal practice and research. Clin Neurophysiol. 2009;120:2008–39.https://doi.org/10.1016/j.clinph.2009.08.016.

5. Chen R, Cros D, Curra A, Di Lazzaro V, Lefaucheur JP, MagistrisMR, et al. The clinical diagnostic utility of transcranial magneticstimulation: report of an IFCN committee. Clin Neurophysiol.2008;119:504–32.

6. Rossini PM, Barker AT, Berardelli A, Caramia MD, Caruso G,Cracco RQ, et al. Non-invasive electrical and magnetic stimulationof the brain, spinal cord and roots: basic principles and proceduresfor routine clinical application. Report of an IFCN committee.Electroencephalogr Clin Neurophysiol. 1994;91:79–92. https://doi.org/10.1016/0013-4694(94)90029-9.

7. Rossini PM, Burke D, Chen R, Cohen LG, Daskalakis Z, Di IorioR, et al. Non-invasive electrical and magnetic stimulation of thebrain, spinal cord, roots and peripheral nerves: basic principlesand procedures for routine clinical and research application: anupdated report from an I.F.C.N. Committee. Clin Neurophysiol.2015;126:1071–107. https://doi.org/10.1016/j.clinph.2015.02.001.

8. Escudero JV, Sancho J, Bautista D, Escudero M, López-Trigo J.Prognostic value of motor evoked potential obtained by transcranialmagnetic brain stimulation in motor function recovery in patientswith acute ischemic stroke. Stroke. 1998;29:1854–9 http://www.ncbi.nlm.nih.gov/pubmed/9731608. Accessed 9 May 2018.

9. Bestmann S, Krakauer JW. The uses and interpretations of themotor-evoked potential for understanding behaviour. Exp BrainRes. 2015;233:679–89. https://doi.org/10.1007/s00221-014-4183-7.

10. Rothwell JC. Techniques and mechanisms of action of transcranialstimulation of the human motor cortex. J Neurosci Methods.1997;74:113–22.

11. Hess CW,Mills KR,Murray NM. Responses in small handmusclesfrommagnetic stimulation of the human brain. J Physiol. 1987;388:397–419 http://www.ncbi.nlm.nih.gov/pubmed/3079553.Accessed 2 Oct 2016.

12. Zhang H, Oweis Y, Mozaffari-Naeini H, Venkatesha S, Thakor NV,Natarajan A. Continuous quantitative motor evoked potentials forspinal cord injury detection. 2nd Int IEEE EMBS Conf Neural Eng.2005;2005:430–3.

13. HérouxME, Taylor JL, Gandevia SC. Correction: the use and abuseof transcranial magnetic stimulation to modulate corticospinal ex-citability in humans. PLoSOne. 2016;11:e0147890. https://doi.org/10.1371/journal.pone.0147890.

14. Grunhaus L, Polak D, Amiaz R, Dannon PN. Motor-evoked poten-tial amplitudes elicited by transcranial magnetic stimulation do not

differentiate between patients and normal controls. Int JNeuropsychopharmacol. 2003;6:371–8.

15. Butler AJ, Kahn S, Wolf SL, Weiss P, Barker A, Beric A, et al.Finger extensor variability in TMS parameters among chronicstroke patients. J Neuroeng Rehabil. 2005;2:10. https://doi.org/10.1186/1743-0003-2-10.

16. Hu Y, Luk KD, LuWW, Holmes A, Leong JC. Prevention of spinalcord injury with time-frequency analysis of evoked potentials: anexperimental study. J Neurol Neurosurg Psychiatry. 2001;71:732–40 http://www.ncbi.nlm.nih.gov/pubmed/11723192. Accessed 16Nov 2016.

17. Braun JC, Hanley DF, Thakor NV. Detection of neurological injuryusing time-frequency analysis of the somatosensory evoked poten-tial. Electroencephalogr Clin Neurophysiol Potentials Sect.1996;100:310–8. https://doi.org/10.1016/0168-5597(96)95115-1.

18. Wang Y, Zhang Z, Li X, Cui H, Xie X, Luk KD-K, et al. Usefulnessof time–frequency patterns of somatosensory evoked potentials inidentification of the location of spinal cord injury. J ClinNeurophysiol. 2015;32:341–5. https://doi.org/10.1097/WNP.0000000000000167.

19. Zhang Z-G, Yang J-L, Chan S-C, Luk KD-K, Hu Y. Time-frequency component analysis of somatosensory evoked potentialsin rats. Biomed Eng Online. 2009;8:4. https://doi.org/10.1186/1475-925X-8-4.

20. Wang Y, Li G, Luk KDK, Hu Y. Component analysis of somato-sensory evoked potentials for identifying spinal cord injury loca-tion. Sci Rep. 2017;7:2351. https://doi.org/10.1038/s41598-017-02555-w.

21. Wang Y, Cui H, Pu J, Luk KDK, Hu Y. Time-frequency patterns ofsomatosensory evoked potentials in predicting the location of spinalcord injury. Neurosci Lett. 2015;603:37–41. https://doi.org/10.1016/J.NEULET.2015.07.002.

22. Hu Y, Luk KDK, Lu WW, Leong JCY. Application of time-frequency analysis to somatosensory evoked potential for intraop-erative spinal cord monitoring. J Neurol Neurosurg Psychiatry.2003;74:82–7. https://doi.org/10.1136/JNNP.74.1.82.

23. Wolf SL, Butler AJ, Campana GI, Parris TA, Struys DM,WeinsteinSR, et al. Intra-subject reliability of parameters contributing to mapsgenerated by transcranial magnetic stimulation in able-bodiedadults. Clin Neurophysiol. 2004;115:1740–7. https://doi.org/10.1016/J.CLINPH.2004.02.027.

24. Lazar RB. Principles of neurologic rehabilitation: McGraw-Hill,Health Professions Division; 1998. https://lib.ugent.be/en/catalog/rug01:000417398. Accessed 10 Jul 2018

25. Grosse P, Cassidy MJ, Brown P. EEG-EMG, MEG-EMG andEMG-EMG frequency analysis: physiological principles and clini-cal applications. Clin Neurophysiol. 2002;113:1523–31.

26. Homan RW, Herman J, Purdy P. Cerebral location of international10–20 system electrode placement. Electroencephalogr ClinNeurophysiol. 1987;66:376–82.

27. Werhahn KJ, Fong JKY, Meyer B-U, Priori A, Rothwell JC, DayBL, et al. The effect of magnetic coil orientation on the latency ofsurface EMG and single motor unit responses in the first dorsalinterosseous muscle. Electroencephalogr Clin NeurophysiolPotentials Sect. 1994;93:138–46.

28. Awiszus F. Chapter 2 TMS and threshold hunting. Suppl ClinNeurophysiol. 2003;56:13–23. https://doi.org/10.1016/S1567-424X(09)70205-3.

29. Kobayashi M, Pascual-Leone A. Transcranial magnetic stimulationin neurology. Lancet Neurol. 2003;2:145–56. https://doi.org/10.1016/S1474-4422(03)00321-1.

30. Day BL, Dressler D, Maertens de Noordhout A, Marsden CD,Nakashima K, Rothwell JC, et al. Electric and magnetic stimulationof human motor cortex: surface EMG and single motor unit re-sponses. J Physiol. 1989;412:449–73. https://doi.org/10.1113/jphysiol.1989.sp017626.

SN Compr. Clin. Med. (2019) 1: –764 780 779

31. Angelova S, Ribagin S, Raikova R, Veneva I. Power frequencyspectrum analysis of surface EMG signals of upper limb musclesduring elbow flexion—a comparison between healthy subjects andstroke survivors. J Electromyogr Kinesiol. 2018;38:7–16. https://doi.org/10.1016/J.JELEKIN.2017.10.013.

32. Henneman E. The size-principle: a deterministic output emergesfrom a set of probabilistic connections. Journal of experimentalbiology. 1985 Mar 1;115(1):105–12.

33. Brown P. Muscle sounds in Parkinson’s disease. Lancet. 1997;349:533–5. https://doi.org/10.1016/S0140-6736(97)80086-4.

34. Lodha N, Chen Y-T, McGuirk T, Fox EJ, Kautz SA, Christou EA,et al. EMG synchrony to assess impaired corticomotor control oflocomotion after stroke. J Electromyogr Kinesiol. 2017;37:35–40.https://doi.org/10.1016/J.JELEKIN.2017.08.007.

35. Fang Y, Daly JJ, Sun J, Hvorat K, Fredrickson E, Pundik S, et al.Functional corticomuscular connection during reaching is weak-ened following stroke. Clin Neurophysiol. 2009;120:994–1002.https://doi.org/10.1016/j.clinph.2009.02.173.

36. Rothwell J, Thompson P, Day B, Boyd S, Marsden C. Stimulationof the humanmotor cortex through the scalp. Exp Physiol. 1991;76:159–200.

Publisher’s Note Springer Nature remains neutral with regard to juris-dictional claims in published maps and institutional affiliations.

SN Compr. Clin. Med. (2019) 1: –764 780780