The usefulness, Neurologia i Neurochirurgia Polska od 2012
[ Pobierz całość w formacie PDF ]
ORIGINAL PAPER/ARTYKU£ ORYGINALNY
The usefulness of accelerometric registration with assessment of tremor
parameters and their symmetry in differential diagnosis of parkinsonian,
essential and cerebellar tremor
Przydatnoœæ rejestracji akcelerometrycznej z ocen¹ parametrów dr¿enia i ich symetrii
w diagnostyce ró¿nicowej dr¿enia parkinsonowskiego, samoistnego i mó¿d¿kowego
Agnieszka Machowska-Majchrzak
1
, Krystyna Pierzcha³a
1
, Stanis³aw Pietraszek
2
, Beata £abuz-Roszak
1
, Wojciech Bartman
1
1
Katedra i Klinika Neurologii, Œl¹ski Uniwersytet Medyczny w Katowicach, Wydzia³ Lekarski z Oddzia³em Lekarsko-Dentystycznym w Zabrzu
2
Instytut Elektroniki, Politechnika Œl¹ska w Gliwicach
Neurologia i Neurochirurgia Polska 2012; 46, 2: 145-156
DOI: 10.5114/ninp.2012.28257
Abstract
Streszczenie
Background and purpose: The aim of the study was to per-
form an analysis of the recorded tremor using accelerometry
and select those parameters that are the most useful in dif-
ferentiation of tremor types.
Material and methods: We examined 45 patients with parkin-
sonian tremor (PT), 39 patients with essential tremor (ET)
and 35 patients with cerebellar tremor (CT). The control
group consisted of 52 healthy persons. The analysis includ-
ed tremor intensity, frequency of spectral peaks, centre fre-
quency, standard deviation of the centre frequency, and har-
monic index. Parameters of tremor were compared between
particular groups of patients with pathological tremor and
with the control group. The side-to-side symmetry of these
parameters was also analysed.
Results: Tremor intensity was significantly higher in patients
than in controls. There was a significant side-to-side asym-
metry of intensity in all patient groups. Significantly lower
peak frequency, centre frequency and standard deviation of
centre frequency were found in patients compared to the con-
trol group. The frequency was symmetric in ET and in con-
trols, but asymmetric in other subjects. The differences bet-
Wstêp i cel pracy: Celem pracy by³a analiza dr¿enia zareje-
strowanego za pomoc¹ akcelerometru z wyodrêbnieniem tych
parametrów, które s¹ najbardziej przydatne w ró¿nicowaniu
dr¿eñ.
Materia³ i metody: Zbadano 45 pacjentów z dr¿eniem par-
kinsonowskim, 39 pacjentów z dr¿eniem samoistnym oraz
35 pacjentów z dr¿eniem mó¿d¿kowym. Grupa kontrolna
sk³ada³a siê z 52 zdrowych neurologicznie osób. Oceniono:
intensywnoœæ dr¿enia, czêstotliwoœæ pików w widmie, czêsto-
tliwoœæ œrodkow¹, odchylenie standardowe od czêstotliwoœci
œrodkowej i indeks harmoniczny. Porównano parametry
dr¿enia miêdzy grupami pacjentów i grup¹ kontroln¹ oraz
grupy pacjentów miêdzy sob¹. Analizowano symetriê bada-
nych parametrów pomiêdzy rêkami.
Wyniki: Intensywnoœæ dr¿enia w grupach chorych by³a zna-
miennie wiêksza w porównaniu z grup¹ kontroln¹. W gru-
pach chorych znamienna by³a asymetria intensywnoœci pomiê-
dzy rêkami. Czêstotliwoœæ pików w widmie, czêstotliwoœæ
œrodkowa oraz odchylenie standardowe od czêstotliwoœci œrod-
kowej by³y znamiennie ni¿sze w grupach chorych w porów-
naniu z grup¹ kontroln¹. Czêstotliwoœæ by³a symetryczna
Correspondence address: Agnieszka Machowska-Majchrzak, Katedra i Klinika Neurologii Œl¹skiego Uniwersytetu Medycznego w Katowicach,
Wydzia³ Lekarski z Oddzia³em Lekarsko-Dentystycznym w Zabrzu, ul. 3-go Maja 13/15, 41-800 Zabrze, e-mail: agnes.majchrzak@vp.pl
Received: 19.03.2011; accepted: 6.09.2011
145
Neurologia i Neurochirurgia Polska 2012; 46, 2
Agnieszka Machowska-Majchrzak, Krystyna Pierzcha³a, Stanis³aw Pietraszek, Beata £abuz-Roszak, Wojciech Bartman
ween hands regarding the standard deviation of centre fre-
quency were significantly greater in all patient groups than
in controls, who revealed no difference of this parameter
between sides. Harmonic index was significantly greater and
asymmetric in all groups of patients when compared to the
control group.
Conclusions: Standard deviation of centre frequency and har-
monic index are the most valuable variables in differentiation
of tremor. The assessment of symmetry of tremor parameters
is useful in discrimination of various types of pathological
tremor.
Key words: tremor, differential diagnosis, quantitative analy-
sis, tremor parameters, power spectrum.
w obu rêkach u osób z dr¿eniem samoistnym i grupie kon-
trolnej, asymetryczna w pozosta³ych grupach. Ró¿nice odchy-
lenia standardowego od czêstotliwoœci œrodkowej pomiêdzy
rêkami by³y znamiennie wiêksze we wszystkich grupach cho-
rych w porównaniu z grup¹ kontroln¹, w której nie stwier-
dza³o siê istotnych ró¿nic pomiêdzy stronami. Znamiennie
wy¿szy by³ indeks harmoniczny, który by³ asymetryczny w obu
rêkach w grupach chorych w porównaniu z grup¹ kontroln¹.
Wnioski: Najwiêksz¹ wartoœæ w ró¿nicowaniu dr¿enia maj¹
odchylenie standardowe od czêstotliwoœci œrodkowej i indeks
harmoniczny. W ró¿nicowaniu typów dr¿enia chorobowego
przydatna jest ocena symetrii badanych wskaŸników.
S³owa kluczowe: dr¿enie, diagnostyka ró¿nicowa, analiza ilo-
œciowa, parametry dr¿enia, widmo mocy.
Introduction
that tremor was not parkinsonian but essential in 26%
of patients diagnosed with PD [3]. Meara
et al.
[4] stud-
ied 404 patients diagnosed with PD and found ET in
12% of those patients. Other studies also confirmed
quite common errors in diagnosis of both these disor-
ders based on the type of the tremor [5-10].
Diagnosis of the tremor according to the clinical
characteristics is sometimes insufficient to establish the
appropriate diagnosis. Methods enabling the quantita-
tive assessment of various tremor parameters and sup-
porting the diagnostic process are therefore important.
Accelerometric registration of tremor provides objec-
tive data on tremor features. We have previously deter-
mined the parameters that are the most useful in differ-
entiation of tremor [11]. The discriminative features
included the standard deviation of centre frequency and
harmonic index; they attested to the rhythmicity and reg-
ularity of tremor.
The aim of the present study was to establish the use-
fulness of spectral analysis parameters along with the
assessment of their symmetry between sides in differ-
ential diagnosis of tremor in patients with clinically diag-
nosed ET, PT or cerebellar tremor (CT).
Differentiation between tremor and other involun-
tary movements is usually simple due to the oscillatory
and rhythmic characteristics of the tremor. Differential
diagnosis of various forms of tremor can be quite chal-
lenging, however, and difficulties encountered might
result in inappropriate therapeutic decisions.
Differentiation between essential tremor (ET) and
parkinsonian tremor (PT) is often problematic because
of the high prevalence of both disorders. Essential
tremor is more common than Parkinson disease (PD),
and the prevalence of both disorders increases with age
of the studied population. It seems therefore that the
proper diagnosis of tremor, especially among elderly sub-
jects, requires considerable insight.
Essential tremor is one of the most common neuro-
logical disorders but its diagnosis may still pose some
difficulties. According to Schrag
et al.
[1] only 50% of
patients diagnosed earlier with ET fulfilled the diag-
nostic criteria. The definition of ET in that particular
study was narrow and included a positive family histo-
ry as a diagnostic criterion. Jain and colleagues verified
the diagnosis of ET in 71 patients and estimated that
the diagnosis was inappropriate in 37% of patients [2].
Repeated evaluation revealed that the most common
proper diagnosis was PD (15%). The authors used the
criteria proposed by the Movement Disorders Society,
which do not consider a positive family history as a pre-
requisite of ET diagnosis; thus, other factors established
during history-taking and neurological examination
influenced the changes in diagnosis.
Diagnostic errors may also occur in the reverse sit-
uation, when ET is diagnosed and treated as PD. One
of the early epidemiological studies on PD estimated
Material and methods
Patients
Our study comprised 119 patients with tremor of
more than 1 year duration diagnosed according to the
Consensus Statement of the Movement Disorders
Society on Tremor [12]. The study group comprised
45 patients with PT, 39 subjects with ET and 35 pa-
tients with CT.
146
Neurologia i Neurochirurgia Polska 2012; 46, 2
The usefulness of accelerometric registration in differential diagnosis of tremor
Exclusion criteria consisted of diagnosed endocrine
dysfunction, use of any medication that might evoke
tremor, use of psychoactive substances, and psychiatric
disorders.
The subgroup of patients with PT comprised 45 pa-
tients (age range 42-81 years; mean 64.2 ± 10.2),
including 17 women (age range 49-81 years; mean
67 ± 10), and 28 men (age range 42-77 years; mean
62.4 ± 10.2). The duration of the disease ranged bet-
ween 1 and 12 years (mean disease duration 5 ± 3 years).
The study involved patients with clinically diagnosed
tremor-dominant or mixed form of idiopathic PD, in
whom uni- or bilateral rest tremor or rest and postural/
kinetic tremor was present. Parkinson disease was diag-
nosed according to the specific diagnostic criteria [13].
Tremor severity was assessed in the ‘off’ stage
on a five-point scale (0-4 pts) according to the third part
of the Unified Parkinson’s Disease Rating Scale
(UPDRS-III), items 20 (rest tremor) and 21 (kinetic
or postural tremor) [14]. Tremor of the first type was
diagnosed in 42 patients, and tremor of the second type
was noted in 3 other patients. Among patients with
tremor of the first type, 15 subjects had an isolated rest
tremor. Rest tremor (UPDRS-III, item 20) had the
range of 1-4 pts (mean: 2.3 ± 0.75 pts) in the hand with
more severe tremor and 0-2 pts (mean: 1.0 ± 1.0 pts)
in the hand with less severe tremor. Postural/kinetic
tremor (UPDRS-III, item 21) had the range of 1-4 pts
(mean: 1.2 ± 1.1 pts) in the hand with more severe
tremor and 0-3 pts (mean: 0.4 ± 0.78 pts) in the hand
with less severe tremor.
The subgroup of patients with ET comprised
39 patients (age range 16-74 years, mean 48.2 ± 17.5)
including 19 women (age range 16-74 years, mean
53.1 ± 18.4) and 20 men (age range 17-69 years, mean
43.6 ± 16). The duration of disease ranged from 2 to
39 years (mean 7.2 ± 7). Essential tremor was diag-
nosed according to the clinical evaluation that revealed
bilateral, more or less symmetrical upper limb tremor of
postural and/or kinetic quality (without rest tremor) with
or without accompanying head tremor, without any
abnormal posturing of the head.
The severity of tremor among patients with ET was
assessed with the scale used in the WHIGET study
(Washington Heights-Inwood Genetic Study of Essen-
tial Tremor) [15]. Thirty-seven patients had postural
and kinetic tremor, while two other subjects had pos-
tural tremor only. The severity of tremor assessed on the
WHIGET scale ranged from 4 to 34 pts (mean 19.66
± 5.65 pts). SevPostural tremor ranged from 1 to 3 pts;
the mean severity of tremor in the hand with more severe
tremor was 1.46 ± 0.64 pts; the mean severity of tremor
in the hand with less severe tremor was 1.46 ± 0.64 pts;
the mean score in both hands was 1.46 ± 0.64 pts.
The severity of kinetic tremor ranged from 0 to
3 pts; the mean severity of tremor in the hand with more
severe tremor was 1.69 ± 0.74 pts; the mean severity of
tremor in the hand with less severe tremor was 1.65 ±
0.73 pts; the mean score in both hands was 1.67 ±
0.74 pts.
The subgroup of patients with CT comprised
35 patients (age range 21-76 years; mean 48 ± 14),
including 19 women (age range 21-69 years; mean 43.8
± 15.5), and 16 men (age range 37-76 years; mean 52.2
± 12.1). The duration of disease ranged from 1 to
37 years (mean 6 ± 7). Cerebellar tremor was diagnosed
according to the clinical evaluation and confirmed with
the presence of a cerebellar lesion in neuroimaging
(computed tomography and/or magnetic resonance
imaging). Patients were qualified for the study if they
had isolated or predominant uni- or bilateral intention-
al tremor with possible associated postural tremor and
without rest tremor.
All 35 patients had some cerebellar signs other than
the tremor. Twelve patients had unilateral cerebellar syn-
drome, while another 23 patients had bilateral symp-
toms and signs.
The severity of tremor among those patients was
assessed with items 11 and 12 of the second part of the
International Cooperative Ataxia Rating Scale (ICARS)
[16]. The severity of intentional tremor in the more affect-
ed limb ranged from 2 to 4 pts (mean 2.7 ± 0.64 pts),
and from 0 to 4 pts (mean 1.1 ± 1.07 pts) in the con-
tralateral limb. The severity of tremor associated with vol-
untary muscle contraction in the more affected limb
ranged from 0 to 4 pts (mean 2.1 ± 1.14 pts), and from
0 to 4 pts (mean 0.94 ± 1.08 pts) in the less affected limb.
The control group consisted of 52 healthy subjects
(age range 16-82 years; mean 52 ± 16.3), including
27 women (age range 16-78 years; mean 50 ± 18) and
25 men (age range 27-82; mean 54 ± 14.5).
Methods
Examinations were performed using a biaxial
accelerometer from Analog Devices with the registra-
tion software from CrossBow. The accelerometer was
mounted at the dorsal surface of the patient’s hand.
Measurements were carried out for the right and left
hand separately and consecutively in three positions:
(1) at rest, when the hand and forearm were fully
147
Neurologia i Neurochirurgia Polska 2012; 46, 2
Agnieszka Machowska-Majchrzak, Krystyna Pierzcha³a, Stanis³aw Pietraszek, Beata £abuz-Roszak, Wojciech Bartman
supported; (2) in the writing position; and (3) when
the upper limb was held outstretched and extended in
pronation.
The registration took 3 minutes (one minute for each
position of the limb). The signal was sampled with the
frequency of 50 Hz per channel and recorded on a com-
puter hard disk. Spectral analysis was done with Mat-
lab software. The initial signal processing included the
restoration of the acceleration vector in the registration
plane and bandpass filtration from 1 to 15 Hz using
a fourth-order Butterworth filter. The distribution of
the spectral density estimator was established through
the calculation of 512-point fast Fourier transformation
(FFT). A Hanning window was used to reduce the spe-
ctral leakage. The spectrum was averaged for 10.2 sec-
ond-wide time epochs of the registered signal.
The analysed parameters included:
• tremor intensity – calculated as the root-mean-square
of acceleration [m/s
2
]; the acceleration is the second
derivative of position with respect to time (m),
• the frequency of peaks within the spectrum [Hz],
• centre frequency – the frequency below which lies
50% of the power in the spectrum and above which
lies the other 50% [Hz],
• standard deviation of centre frequency or dispersion
about median frequency – the frequency width of an
interval around the centre frequency that contains
68% of the total power in the spectrum; it reflects the
degree of discoordination of the tremor; a very rhyth-
mic tremor has a small value of that variable, indicat-
ing that most of the energy is produced within a nar-
row frequency band [Hz],
• harmonic index (HI) or the index of harmonic con-
tents in the spectrum. It defines the distance of the
spectrum to the single narrow peak; that value is nor-
malized to the highest peak.
The obtained results were statistically analysed using
STATISTICA v.7.1 software.
The basic statistical measures, including mean, medi-
an, maximum, minimum, standard deviation, skewness,
and kurtosis, were calculated for all interval variables.
The distribution of the variables and its compatibility
with normal distribution was tested with the Shapiro-
Wilk test. The association of tremor parameters with age
in patients was tested with Spearman rank correlation
coefficient. The differences in tremor parameters among
patient subgroups were tested with Kruskal-Wallis rank
analysis of variance, supplemented with post-hoc analy-
sis with the Tukey test. The Tukey test (in contrast to the
Mann-Whitney
U
-test) takes into account the ‘cumula-
tive alpha effect’ due to the multiple comparisons.
Analysis of variance was performed for the variables
standardized according to age. The results of that analy-
sis account for the differences in age among studied
groups (the impact of age was reduced due to the stan-
dardization). The Wilcoxon signed-rank test was used
to compare the tremor parameters between hands with
greater and smaller intensity of tremor. This test was
used instead of Student’s
t
-test for paired samples
because of non-normal distribution and high absolute
values of skewness and kurtosis for multiple variables.
The relationship between intensity and frequency of
tremor was verified with linear regression.
P
-values
< 0.05 were considered statistically significant.
Tremor parameters were compared between groups
with pathological tremor and the control group as well
as among subgroups with particular tremor. The fol-
lowing variables were analysed: (a) mean values from
both hands; (b) data grouped according to the accelerom-
etry results from the hand with greater and smaller inten-
sity of tremor; (c) differences in values of particular para-
meters between sides. The values of studied parameters
were also compared between hands with greater and
smaller intensity of tremor. The relationship between
intensity and frequency of tremor was also analysed.
Results
Tremor intensity
Analysis of data derived from the more affected hand
and from the mean values of both hands revealed that
the tremor intensity was significantly greater in any type
of pathological tremor than in controls (
p
< 0.001 for
each difference between ET, PT or CT subgroups and
controls). Tremor intensity was greater in ET than in
PT or CT (
p
< 0.001). Tremor intensity was similar in
patients with PT and CT.
Patients with ET (
p
< 0.001) or CT (
p
= 0.04) had
greater tremor intensity than controls also in the less
affected hand. In patients with PT, tremor intensity in
the less affected hand was similar to healthy controls.
The difference of tremor intensity between sides was
also significantly greater in patients than in controls
(PT vs. controls,
p
= 0.005; ET vs. controls,
p
< 0.001,
CT vs. controls,
p
=0.02); it was also greater in ET
patients when compared with other patients (
p
< 0.001).
Tremor intensity in the hand with more severe tremor
was greater than in the hand with less severe tremor
(
p
< 0.001). There was no difference between the hand
148
Neurologia i Neurochirurgia Polska 2012; 46, 2
The usefulness of accelerometric registration in differential diagnosis of tremor
with greater and that with lesser tremor intensity in the
control group (Tables 1 and 2).
all three studied subgroups of patients was lower than
in controls, both in the hand with greater tremor inten-
sity and in the hand with smaller tremor intensity.
The frequencies of peaks in the more affected hand
were also significantly different among groups (CT <
PT < ET) (
p
< 0.001), while in the less affected hand
the frequency of peaks was similar in PT and ET, but
lower in CT than in other patient subgroups (
p
< 0.001).
The difference in frequency of peaks between sides
was significantly greater in PT than in controls or other
subgroups of patients (
p
< 0.001); the two other patient
Frequency of peaks within the spectrum
The frequency of peaks was significantly lower in
patients than in controls (
p
< 0.001) when mean values
from both hands in particular subgroups of patients were
analysed. This frequency was lower in PT than in ET
but it was the lowest in CT. Results in particular groups
were significant (
p
< 0.001). The frequency of peaks in
Table 1. Results of accelerometry – tremor parameters in patients with parkinsonian tremor (PT), essential tremor (ET), cerebellar tremor (CT) and control group
Tremor parameters
More trembling hand;
Less trembling hand;
Both hands
Odds of values
median (range)
median (range)
(median)
between hands;
median (range)
Tremor intensity [m/s
2
]
PT
0.32 (0.07-3.17)
a,b
0.11 (0.04-1.33)
b
0.16
a,b
0.19 (0.01-2.02)
a,b
1.33 (0.14-9.42)
a,c,d
0.52 (0.08-3.61)
a,c,d
0.88
a,c,d
0.79 (0.01-5.81)
a,c,d
ET
0.32 (0.06-4.08)
a,b
0.14 (0.05-1.67)
a,b
0.22
a,b
0.16 (0.0-2.93)
a,b
CT
0.095 (0.017-0.213)
c,b,d
0.088 (0.016-0.204)
b,d
0.091
c,b,d
0.006 (0.0-0.19)
c,b,d
Controls
Frequency of peak [Hz]
PT
5.18 (4.10-6.64)
a,b,d
5.97 (4.10-9.77)
a,d
5.57
a,b,d
0.78 (0.09-4.29)
a,b,d
6.54 (4.20-11.70)
a,c,d
6.64 (4.10-11.10)
a,d
6.64
a,c,d
0.30 (0.0-1.23)
c
ET
3.13 (1.96-5.01)
a,c,b
3.52 (1.95-4.92)
a,b,c
3.40
a,c,b
0.32 (0.0-2.20)
c
CT
8.69 (4.79-12.01)
c,b,d
8.89 (4.93-11.70)
c,b,d
8.80
c,b,d
0.471 (0.02-1.66)
c
Controls
Centre frequency [Hz]
PT
5.60 (4.19-6.80)
a,b,d
6.50 (4.59-8.70)
a,d
5.93
a,b,d
0.87 (0.02-2.76)
a,b,d
6.64 (4.36-11.10)
a,c,d
6.65 (4.31-10.60)
a,d
6.65
a,c,d
0.25 (0.01-0.92)
c
ET
3.18 (1.94-4.89)
a,c,b
3.61 (1.96-4.50)
a,b,c
3.36
a,c,b
0.21 (0.01-1.99)
c
CT
8.37 (7.11-10.54)
c,b,d
8.39 (7.18-10.74)
c,b,d
8.37
c,b,d
0.198 (0.0-0.72)
c
Controls
SD of centre frequency [Hz]
PT
1.17 (0.10-3.71)
a,d
2.54 (0.10-4.98)
a,d
1.95
a,d
0.88 (0.0-2.74)
a,b,d
1.76 (0.20-3.61)
a,d
2.35 (0.78-3.67)
a,d
2.05
a,d
0.47 (0.0-1.66)
a,c
ET
0.98 (0.10-1.66)
a,c,b
1.37 (0.29-2.44)
a,b,c
1.17
a,c,b
0.49 (0.10-1.95)
c
CT
4.54 (3.42-5.18)
c,b,d
4.64 (3.22-5.27)
c,b,d
4.59
c,b,d
0.195 (0.0-1.08)
c,b
Controls
Harmonic index [Hz]
PT
0.95 (0.90-0.99)
a
0.90 (0.81-0.99)
a
0.94
a
0.03 (0.0-0.15)
b,d
0.95 (0.88-0.98)
a
0.93 (0.89-0.98)
a
0.94
a
0.02 (0.0-0.6)
c
ET
0.95 (0.9-0.99)
a
0.93 (0.88-0.98)
a
0.94
a
0.02 (0.0-0.07)
c
CT
0.751 (0.62-0.88)
c,b,d
0.732 (0.634-0.86)
c,b,d
0.746
c,b,d
Controls
0.030 (0.0-0.134)
SD – standard deviation;
a
significant difference between patients and control group;
b
significant difference in comparison with ET;
c
significant difference in comparison with PT;
d
significant difference in comparison with CT; (p < 0.05, Kruskal-Wallis one-way analysis of variance)
149
Neurologia i Neurochirurgia Polska 2012; 46, 2
[ Pobierz całość w formacie PDF ]