On the Impact of Pathological Tremor Intensity on Noninvasive

Characterization of Motor Unit Discharge Properties

Petra Povalej Bržan

, Vojko Glaser

, Simon Zelič

, Juan Álvaro Gallego

Juan Pablo Romero Muñoz

3,4

and Aleš Holobar

System Software Laboratory, University of Maribor, Smetanova ulica 17, 2000 Maribor, Slovenia

Bioengineering Group, Consejo Superior de Investigaciones Científicas, Madrid, Spain

Neurology Department, 12 de Octubre University Hospital, Madrid, Spain

Medical School, Faculty of Health Sciences, Francisco de Vitoria University, Madrid, Spain

Keywords: Surface Electromyograms, Decomposition, Motor Units, Pathological Tremor, Motor Unit Discharge

Patterns.

Abstract: The impact of severity of pathological tremor on surface EMG decomposition was systematically assessed

on eight essential tremor patients. The inertial data and surface EMG signals were concurrently recorded

from wrist extensor and flexor muscles of both patients’ arms. The inertial recordings were segmented into

different tremor cycles and the tremor amplitude was assessed in each tremor cycle. Surface EMG was

decomposed by Convolution Kernel Compensation (CKC) technique in order to yield individual motor unit

discharge patterns in each tremor cycle. Accuracy of EMG decomposition was assessed for each identified

motor unit and was largely uncorrelated with tremor amplitude. In all the patients, the percentage of EMG

energy identified by decomposition and the number of identified motor units were found to be positively

correlated with tremor amplitude, though the correlation was relatively weak and not always significant.

The results demonstrate that the CKC decomposition not only copes with moderate and severe tremor but

also improves its performance with tremor intensity.

1 INTRODUCTION

Pathological tremor is one of the most common

movement disorders, affecting up to 15% of

population aged between 50 and 89 years (Benito-

León et al., 2006). Although not life-threatening,

tremor causes serious difficulties in activities of

daily living and greatly reduces the quality of life of

affected person.

Among pathological tremors, essential (ET) and

Parkinsonian (PD) tremors are the most common.

None of them is completely understood and their

origin is largely unknown. It has been previously

demonstrated that in both tremors peripheral (PNS)

and central nervous system (CNS) interact with each

other (Halliday et al., 1995) to produce rhythmic,

involuntary oscillatory movements of various body

parts.

Involvement of CNS into the tremor genesis is

typically assessed by neuroimaging techniques, such

as Dynamic imaging of Coherent Sources (Gross et

al., 2001) and Renormalized partial directed

coherence (Scheltera et al., 2009) that offer an

important insight into the connectivity and

tremorogenic activity of central networks, such as

primary sensor motor cortex, thalamus and basal

ganglia. For example, it has been demonstrated that

in PD patients basic tremor frequency and its higher

harmonics could have different cortical origins

(Muthuraman et al., 2012). The same differences

have been also observed in ET patients, but to much

lesser extent than in PD (Muthuraman et al., 2012).

On the other hand the tremorogenic activity of

PNS is measured by the inertial and EMG sensors.

Inertial recordings are easy to collect and, thus, very

appealing candidates for the tremor quantification,

diagnosis and/or tracking (Deuschl et al., 1995);

(Muthuraman et al., 2011); (Saunders-Pullman et al.,

2008); (Groznik et al., 2013). However, although

agreeing on the diagnostic power of the inertial

recordings, the current studies offer limited insight

into the origin of mechanic oscillations. The latter

can be better assessed by the EMG measurements.

However, practically all published studies focus on a

126

Povalej Bržan P., Glaser V., Zeli

c S., Gallego J., Romero Muñoz J. and Holobar A..

On the Impact of Pathological Tremor Intensity on Noninvasive Characterization of Motor Unit Discharge Properties.

DOI: 10.5220/0004664001260132

In Proceedings of the International Congress on Neurotechnology, Electronics and Informatics (DeNeuro-2013), pages 126-132

ISBN: 978-989-8565-80-8

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

very simple surface EMG metrics, such as EMG

rectification, to estimate the neural drive to

tremorogenic muscles (Gross et al., 2001);

(Scheltera et al., 2009); (Raethjen et al., 2009).

Moreover, surface EMG acquisition is typically

limited to bipolar recording systems that offer no

insight into discharges of individual motor units

(MUs). This is a problematic step as surface EMG

amplitude is known to reflect many anatomical

properties of investigated muscles that interfere with

the central control component (voluntary and

tremorogenic neural drive) from spinal and

supraspinal neural circuits (Farina et al.; 2008,

2010).

A very few studies of individual MU discharge

characteristics in pathological tremor exist. They

focus mostly on PD patients (Das Gupta, 1963);

(Dietz et al., 1974); (Dengler et al., 1986);

(Christakos et al., 2009), whereas the studies of MUs

in ET patients are close to absent. Moreover, all

these studies rely on invasive indwelling EMG

recordings that limit the number of concurrently

identified MU to a very few and hinder their long-

term tracking.

Recently, advanced surface EMG decomposition

technique has been demonstrated to identify the

discharge patterns of up to several tens individual

motor units in wrist flexors and extensors of tremor

patients (Holobar et al., 2012). The work presented

herein extends the one in (Holobar et al., 2012) by

systematically assessing the relationship between the

identified MU discharge patterns and the severity of

pathological tremor. This study focuses on ET

patients, whereas PD patients will be addressed in

the future work.

2 PROCEDURE

Eight ET patients (3 females, 5 males, age of 688

years), with mild (4 patients), moderate (2 patients)

and severe (2 patients) tremor according to the Fahn-

Tolosa-Marin scale participated to the experiment.

Six patients had family history of essential tremor.

Three patients (patient B, C and D) were taking

medications for the treatment of ET syndromes.

Detailed description of the patients is presented in

Table 1.

The experiments were conducted in accordance

with the Declaration of Helsinki and approved by

the ethics committee of the 12 de Octubre University

Hospital, Madrid, Spain. The patients received a

detailed explanation of the study and gave written

informed consent prior to participation.

Table 1: Description of patients involved in the study.

Patient Gender Age Family history

Age at tremor

onset

Tremor

severity

A F 69 yes 48 severe

B F 63 yes 56 moderate

C M 76 no 74 moderate

D M 67 yes 46 mild

E M 75 yes 32 mild

F M 69 yes 59 mild

G F 51 no 41 mild

H M 73 yes 50 severe

2.1 Experimental Protocol

The patients were equipped with surface EMG

sensors and inertial measuring units (IMUs)

mounted bilaterally over both upper extremities.

Inertial sensors with 3D accelerometer, gyroscope

and magnetometer (Technaid Motion Capture

System) have been mounted to the third metacarpal,

edge of the forearm (dorsal side), olecranon process

and distal part (dorsal side) of each arm. The inertial

signals were sampled at 200 samples/s and recorded

concurrently with surface EMG.

Matrices of 12×5 surface electrodes (OT

Bioelettronica and LISiN, Italy) were centred over

the bellies of the left and right flexor carpi radialis

and extensor carpi ulnaris. Before mounting the

matrices, the skin has been lightly abraded by using

the abrasive paste (Meditec–Every, Parma, Italy)

and cleansed afterward. Electrical conductivity was

ensured by filling the electrode grids with

conductive gel (Meditec–Every, Parma, Italy). The

signals were amplified, band-pass filtered (3 dB, 10-

500 Hz) and sampled at 2048 samples/s (EMG-

USB2 amplifier, OT Bioelettronica).

Patient performed three repetitions of the

following tremor-triggering tasks, each in duration

of at least 30 s:

1. Arms outstretched against gravity (AO): after 5

seconds of rest, the patient outstretched his/her

arms and maintained the posture against gravity.

2. Arms outstretched against gravity with weights

(WO): the same as AO task with additional

weight load of ~1kg applied to both hands.

3. Arms supported + postural tremor elicited (PO):

patient was sitting relaxed, with arms supported

in elbow and with wrists hold extended against

the gravity.

OntheImpactofPathologicalTremorIntensityonNoninvasiveCharacterizationofMotorUnitDischargeProperties

127

2.2 Data Analysis

The acquired surface EMG signals were

decomposed by Convolution Kernel Compensation

(CKC) technique (Holobar et al., 2012), which is

fully automatic and provides so called Pulse-to-

Noise ratio (PNR) which has been demonstrated to

reliably assess the decomposition accuracy for every

identified motor unit (Holobar et al., 2013). Only

MUs, with PNR > 26 dB and, thus, sensitivity in

identification of discharges > 80 % and false alarm

rates < 5 % (Holobar et al., 2013) were kept for

further analysis. All the remaining motor units were

discarded.

The surface EMG signals were reconstructed

using the MU action potential (MUAP) shapes

estimated by spike-triggered averaging of surface

EMG (Holobar et al., 2010) and the discharge times

estimated from the surface EMG decomposition.

The surface EMG reconstructed in this way was

compared with the original signal by the following

signal-to-interference ratio (SIR):



(() ())

( ) 1 100 %

()

iij

xn z n

SIR i



 







 





(1)

where x

(n) denotes the i-th surface EMG channel

and z

(n) stands for the MUAP train of the j-th MU

as reconstructed from the i-th surface EMG signal.

In addition to decomposition, surface EMG was

also band-pass filtered (50-200 Hz) and full-wave

rectified (Halliday et al., 1995). Relative power

(RP

) of basic tremor frequency was then

calculated for each EMG channel as the ratio

between the peak in the power spectrum of rectified

EMG and its total power.

The recorded IMU data has been upsampled to

2048 samples/s, synchronized with EMG signals and

processed by Ensemble Empirical Mode

Decomposition (Wu et al. 2009) in order to extract

tremor component from voluntary movement

(Rocon et al., 2006). The extracted tremor

component and the MU discharge patterns have been

segmented into different tremor cycles (Figure 5).

The following metrics have been calculated for each

identified tremor cycle: the mean tremor amplitude

and power as assessed from inertial data, the number

of active MUs as assessed from surface EMG

decomposition, the total number of MU discharges

and standard deviation (SD) of their discharge times

within the tremor cycle.

All extracted features were statistically analysed

in RStudio software. The Wilcoxon matched pairs

signed rank test was used to compare the extracted

features between extensors and flexors muscles and

Spearman correlation coefficient was used to

evaluate the relationship between the RP

and SIR

and inertial tremor measurements and MU discharge

characteristics, respectively. Statistical significance

was set to P<0.05. The results are presented as mean

± SD.

3 RESULTS

From 2 to 30 (19.2 ± 8.8) MUs were identified in

extensors muscles and from 0 to 32 MUs (18.9 ±

11.1) in flexor muscles in all the patients and all the

tasks. All the cases with zero identified MUs

coincided with the lack of muscle activity as

measured by surface EMG. PNR was not

significantly correlated with RP

, except in patient

H, where a positive correlation of 0.48 and 0.61 was

observed in the extensor and flexor muscles of

dominant tremor arm, respectively.

Figures 1 and 2 depict the distribution of the

maximal RP

(maximum calculated across all EMG

channels per matrix) and the SIR, as defined in Eq.

(1), calculated on the channel with maximal RP

(SIR*). Across the patients, the maximal RP

varied from 0.04 to 0.72 (0.24 ± 0.17) on extensor

and from 0.05 to 0.65 (0.24 ± 0.17) on flexor

muscles (Figure 1).

Figure 1: Distribution of the maximal RP

across all the

tasks of individual patients.

NEUROTECHNIX2013-InternationalCongressonNeurotechnology,ElectronicsandInformatics

128

Figure 2: Distribution of SIR* across all the task

repetitions of individual patient. In each task repetition,

the SIR* was calculated on the EMG channel with

maximal RP

Wilcoxon matched pairs signed rank test

indicated no significant differences in maximal RP

of extensors and flexors muscles, except in patient B

where maximal RP

was significantly higher in

extensors (z = -2.72). In EMG channel with maximal

, the SIR* varied from 2 to 66% (39.6 ± 14.2 )

on extensor and from 0 to 68% (31.9 ± 15.2) on

flexors muscles. Wilcoxon matched pairs signed

rank test indicated significant difference between

SIR of extensor and flexor muscles in 4 out of 8

patients (Patient A, F, G and H in Figure 2).

The relation between the maximal RP

and the

corresponding SIR in each task repetition is depicted

in Figures 3 and 4 and summarized in Table 2.

Practically all the patients exhibited a positive

correlation between the RP

and the SIR* in at least

one out of extensor and flexor muscles. In six out of

eight patients, the correlation was statistically

significant (P < 0.05). In patients G and H the

correlation was not statistically significant. In patient

H correlation was negative but SIR* was extremely

high over all measurements on wrist extensors

(Figure 4) and flexor muscles (results not shown).

In patient G, the tremor amplitude was always low

(Figure 1).

Figure 3: The relationship between the RP

of wrist

extensor muscle and the SIR* (patient E). Blue plusses

depict results on dominant hand, the pink dots represent a

non-dominant hand.

Figure 4: The relationship between the RP

of wrist

extensor muscle and the SIR* in patient H. Blue plusses

depict results on dominant hand, the pink dots represent a

non-dominant hand.

Table 2: Correlation coefficients between the maximum

and SIR* in both arms.

▲

statistically significant

correlation, Spearman Correlation test (P < 0.05).

Patient

Extensors

& SIR*

Flexors RP

& SIR*

Extensors No.

MUs & SIR

Flexors No.

MUs & SIR

A 0.49

▲

0.16 0.79

▲

0.57

▲

B 0.51

▲

0.42 0.76

▲

0.87

▲

C 0.62

▲

0.19

▲

0.71

▲

0.29

D 0.30 0.80

▲

0.93

▲

0.80

▲

E 0.41 0.74

▲

0.71

▲

0.56

▲

F -0.25 0.54

▲

0.36 0.35

G 0.13 0.05 0.37 0.73

▲

H - 0.33 0.36 0.00 0.89

▲

In dominant tremor arm alone a significant

correlation between the RP

and the SIR* was

OntheImpactofPathologicalTremorIntensityonNoninvasiveCharacterizationofMotorUnitDischargeProperties

129

observed in the wrist flexor or extensor muscles in 4

patients only. The reduction in sample size (3 AO, 3

PO and 3 WE taks) is the most likely reason. In the

nondominant tremor arm the correlation was

frequently not significant, likely because of

relatively low tremor amplitude values, especially in

patients with unilateral tremor (Figure 3).

The discharge patterns of individual MUs as

identified by surface EMG decomposition of

extensor and flexor muscles of dominant tremor

hand in patient H are depicted in Figure 5, along

with the tremor component as recorded by inertial

sensor placed at the patient’s wrist. Out-of-phase

flexor-extensor activity is clearly visible. The MU

discharge patterns in the same patient but during the

period of low tremor amplitude are depicted in

Figure 6. The difference in MU discharge patterns

between the periods with high and low tremor

amplitudes are clearly visible.

Figure 5: discharge patterns of individual MUs identified

by surface EMG decomposition of extensor and flexor

muscles of dominant tremor arm in patient E during the

period of high tremor amplitude. Each filled rectangle

denotes one MU firing. Thin black line depicts the tremor

component as measured by the inertial sensor. Circles

denote the segmentation of inertial tremor recording into

the tremor cycles.

Figure 6: discharge patterns of individual MUs identified

by surface EMG decomposition of extensor and flexor

muscles of dominant tremor hand in patient E during the

period of low tremor amplitude.

The global relationship between the tremor

amplitude as assessed by inertial measurements and

the identified MU discharge characteristics in each

tremor cycle are exemplified in Figure 7 and

summarized in Table 3. In the tremor dominant arm,

all the patients exhibited relatively weak but

statistically significant positive correlation between

the tremor amplitude and the number of MUs

discharges per tremor cycle in at least one of

measured muscles (wrist extensors and flexors). The

same applies for the number of active MUs (Table

3).

Figure 7: The number of MUs discharges per tremor cycle

as identified from the left wrist extensor of patient H (left)

and the dispersion of their discharge times (right) vs.

tremor amplitude in each tremor cycle. Results are

accumulated over three repetitions of AO, WE and PO

tasks.

Table 3: Correlation coefficients between the number of

MUs (No. MUs), the number of their discharges (No. MUs

disch.), and their dispersion (SD MUs) in individual

tremor cycle and the tremor amplitude as measured by

inertial sensor placed at dominant wrist; n.s. - not

significant.

Patient

Flexors Extensors

No.

MUs.

No.

MUs

disch.

SD MUs

No.

MUs.

No.

MUs

disch

SD MUs

A n.s. n.s. -0.25 0.24 0.24

-0.48

B 0.19 0.21 n.s. n.s. n.s. -0.20

0.47

0.25 n.s. 0.19 0.16 -0.26

0.47 0.46

-0.18 n.s n.s -0.14

E 0.26 0.29 n.s.

0.42

0.32

-0.43

F n.s. 0.20 n.s. 0.25 0.31 n.s.

G 0.11 0.12 n.s. 0.20 0.18 -0.25

0.53 0.52

n.s. 0.37

0.45 -0.44

Dispersion of MU discharges within the tremor

cycles, as measured by SD of MU discharge times

was inversely related to tremor amplitude, though

also in this case the correlation coefficients were

relatively small.

It is noteworthy that the strongest correlations

between the tremor amplitude and investigated MU

discharge characteristics were observed in patients

with strongest tremor (A, C, D, E and H), whereas

they were much weaker or even not significant in the

100ms

1m/s

extensorflexor

100ms

1m/s

extensorflexor

0 1 2 3 4

Tremor amplitude (m/s)

No. MU discharges

0 1 2 3 4

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

SD of MU discharge times (s)

Tremor amplitude (m/s)

NEUROTECHNIX2013-InternationalCongressonNeurotechnology,ElectronicsandInformatics

130

patients B, F and G with relatively mild tremor. The

strongest correlations were observed in patient with

the most severe tremor (patient H).

Also interesting, there were frequently different

trends in MU behaviour with increasing tremor

amplitude. Patient A, for example, increased the

number of active MUs in extensors and

synchronized their discharges. The same MU

synchronization was also observed in flexor muscle,

though to much less extent. Patient D, on the other

hand, mostly increased the number of active MUs

and their discharges in the flexor muscle.

4 CONCLUSIONS

In this study, the CKC-based decomposition of

multichannel surface EMG was tested on eight ET

patients with different tremor severity. Two different

measures of decomposition performance have been

applied. First, the PNR measure of decomposition

accuracy (Holobar et al., 2013) was found to be

largely uncorrelated with tremor severity, except in

patient H where correlation was positive for both

extensor and flexor muscles. Note that the higher the

PNR, the more accurate the decomposition and that

MUs with PNR > 30 dB typically exhibit sensitivity

in the identification of their discharges greater than

90% (Holobar et al., 2013).

The second measure of decomposition

performance was SIR as defined in Eq. (1). SIR

measures the relative proportion of signal energy

that is accounted for by identified MUs and typically

ranges between 20 % and 50 % in healthy subjects

(Holobar et al., 2010). The values measured in ET

patients are in perfect agreement with these values.

No negative impact of pathological tremor on SIR

was observed. On the contrary, as demonstrated by

results in Figures 3 and 4 and Table 2, the SIR was

mostly positively correlated with the tremor

intensity as measured by RP

i.e. the relative power

of rectified surface EMG at basic tremor frequency.

Not surprisingly, the SIR was also positively

correlated with the number of identified MUs (Table

2).

Detailed analysis of MU discharge patterns in

each tremor cycle revealed weak but consistent

positive correlations between the number of MUs

discharges and MU synchronisation and tremor

amplitudes as measured by inertial sensors. The

observed weakness of correlation can be contributed

to many different factors. First, the inertial sensors

measure the net mechanical oscillations contributed

by many different muscles. In our case, not all the

muscles of wrist were measured, hindering the full

interpretation of the mechanical oscillations with the

introduced electrophysiological variables. Second,

relatively simple and crude metrics of MU

synchronization and neural drive to the muscle were

used in this study, preferring, for example, the time

resolution over the accurate estimation of MU

synchronization. Third, we did not quantify the

extent of out-of-phase flexor-extensor activity

depicted in Figure 5. All this factors likely

contributed to conservative estimation of relation

between the observed MU discharge properties and

kinetic tremor properties and need to be addressed in

the future work.

In conclusion, the impact of tremor severity on

surface EMG decomposition has been systematically

assessed on eight ET patients. As demonstrated by

the results, the CKC decomposition not only fully

copes with severe tremor but also improves its

performance in sense of percentage of identified

signal energy and the number of identified motor

units. This makes it an appealing novel tool for non-

invasive and long-term tracking and physiological

interpretation of pathological tremor.

ACKNOWLEDGEMENTS

This study was supported by the Commission of the

European Union, within Framework 7, under Grant

Agreement number ICT-2011.5.1-287739

"NeuroTREMOR: A novel concept for support to

diagnosis and remote management of tremor".

REFERENCES

Benito-León, J., Louis, E. D., 2006. Essential tremor:

emerging views of a common disorder. In Nature

Clinical Practice Neurology 2. pp. 666-678.

Christakos, C. N., Erimaki, S., Anagnostou, E.,

Anastasopoulos, D., 2009. Tremor-related motor unit

firing in Parkinson’s disease: implications for tremor

genesis. In J Physiol. vol. 587. pp. 4811-4827.

Das Gupta, A., 1963 Paired response of motor units during

voluntary contraction in Parkinsonism. In J Neurol

Neurosurg Psychiatry. vol. 26(3). pp. 265–268.

Dengler, R., Wolf, W., Schubert, M., Struppler, A., 1986

Discharge pattern of single motor units in basal

ganglia disorders. In Neurology. vol 36(8). pp. 1061-

1066.

Deuschl, G., Lauk, M., Timmer, J., 1995. Tremor

classiﬁcation and tremor time series analysis. In

Chaos. vol. 5. pp. 48–51.

Dietz, V., Hillesheimer, W., Freund, H. J., 1974.

OntheImpactofPathologicalTremorIntensityonNoninvasiveCharacterizationofMotorUnitDischargeProperties

131

Correlation between tremor, voluntary contraction,

and firing pattern of motor units in Parkinson's

disease. In J Neurol Neurosurg Psychiatry. vol. 37(8).

pp. 927–937.

Farina, D., Holobar, A., Merletti, R., Enoka R.M., 2010

Decoding the neural drive to muscles from the surface

electromyogram. In Clin Neurophysiol. vol. 121. pp.

1616–1623.

Farina, D., Negro, F., Gazzoni, M., Enoka, R.M., 2008

Detecting the unique representation of motor-unit

action potentials in the surface electromyogram. In J.

Neurophysiol. vol. 100. pp. 1223–1233.

Gross, J., Kujala, J., Hamalainen, M., Timmermann, L.,

Schnitzler, A., Salmelin, R., 2001. Dynamic imaging

of coherent sources: Studying neural interactions in

the human brain. In Proc. Natl. Acad. Sci. USA 98. pp.

694-699.

Groznik, V. , Guid, M., Sadikov, A., Možina, M.,

Georgiev, D., Kragelj, V., Ribarič, S., Pirtošek, Z.,

Bratko, I., 2013. Elicitation of neurological knowledge

with argument-based machine learning. In Artificial

Intelligence in Medicine. vol. 57(2). pp. 133-44.

Halliday, D. M., Rosenberg, J. R., Amjad, A. M., Breeze,

P., Conway, B. A., Farmer, S. F., 1995. A framework

for the analysis of mixed time series/point process data

- theory and application to the study of physiological

tremor, single motor unit discharges and

electromyograms. In Prog Biophys Mol Biol. vol.

64, no. 2/3, pp. 237-278.

Holobar, A., Glaser, V., Gallego, J.A., Dideriksen, J.L.,

Farina, D., 2012. Non-invasive characterization of

motor unit behaviour in pathological tremor. In J

Neural Eng. vol. 9.

Holobar, A., Minetto, M. A., Botter, A., Negro, F., Farina

D., 2010. Experimental analysis of accuracy in the

identification of motor unit spike trains from high-

density surface EMG. In IEEE trans. neural syst.

rehabil. eng.vol. 18. no. 3. pp. 221-229.

Holobar, A., Minetto, M. A., Farina, D., 2013. A signal-

based approach for assessing the accuracy of high-

density surface EMG decomposition. Submitted to 6th

International IEEE/EMBS Conference on Neural

Engineering 2013.

Muthuraman, M., Heute, U., Arning, K., Anwar, A.R.,

Elble, R., Deuschl, G., Raethjen, J., 2012. Oscillating

central motor networks in pathological tremors and

voluntary movements. What makes the difference?. In

NeuroImage. vol. 60. pp. 1331–1339.

Muthuraman, M., Hossen, A., Heute, U., Deuschl, G.,

Raethjen, J., 2011. A New Diagnostic Test to

Distinguish Tremulous Parkinson’s Disease from

Advanced Essential Tremor. In Movement Disorders.

vol. 26. no. 8. pp. 1548-1552.

Raethjen, J., Govindan, R.B., Muthuraman, M., Kopper,

F., Volkmann, J., Deuschl, G., 2009. Cortical

correlates of the basic and ﬁrst harmonic frequency of

Parkinsonian tremor. In Clin Neurophysiol. vol. 120.

pp.1866–1872.

Rocon, E., Andrade, A.O., Pons, J. L., Kyberd, P., Nasuto,

S. J., 2006. Empirical mode decomposition: a novel

technique for the study of tremor time series. In Med.

Biol. Engineering and Computing. vol. 44(7). pp. 569-

582 .

Saunders-Pullman, R., Derby, C., Stanley, K., Floyd, A.,

Bressman, S., Lipton, R.B., Deligtisch, A., Severt, L.,

Yu, Q., Kurtis, M., Pullman, S.L., 2008. Validity of

spiral analysis in early Parkinson's disease, In Mov

Disord. vol. (4). pp. 531-537.

Scheltera, B., Timmer, J., Eichler, M., 2009. Assessing the

strength of directed inﬂuences among neural signals

using renormalized partial directed coherence. In

Journal of Neuroscience Methods. vol. 179 (1). pp.

121–130.

Wu, Z., Huang, N.E., 2009. Ensemble Empirical Mode

Decomposition: a Noise-assisted Data Analysis

Method. In Adv Adapt Data Anal. vol. 1. pp. 1–41.

NEUROTECHNIX2013-InternationalCongressonNeurotechnology,ElectronicsandInformatics

132