A Study for Automatic Diagnosing System of Parkinson Disease

A Systematic Analysis of Parkinsonian Tremor by Accelerometer

Ichiro Fukumoto

Institute of Biological Engineering, Nagaoka University of Technology, Nagaoka city, Niigata, Japan

Keywords: Parkinson Disease, Physiological Diagnosis, Biofeedback, Accelero-Meter, Main Frequency.

Abstract: New objective diagnosing methods of Parkinson Disease is proposed with 3 D accelerometers. A

mathematical model based on the peripheral feedback theory is tested by computer simulation with good

coincident with the clinical data. We have found the main frequency of Parkinsonian tremor is about 4Hz in

arms comparing the one of physiological tremor (8Hz). Patients in L-Dopa treatment have been measured

by the system that corresponds well to a parameter F

gk

that indicates the fatigue of intrafusal muscle.

Biofeedback training by sound and visual parameters are also proved with good tremor improvement in its

main frequency and severity.

1 INTRODUCTION

The Parkinson’s disease (morbidity 0.1%) is

etiology unknown neurologic disease that makes

adults unable to work with three main symptoms

(tremor, rigidity, bradykinesia). The problem of the

diagnosis depends upon the neurologists’ subjective

decision. We have been trying to help non-specialist

using an automatic diagnosing unit with 3D

accelerometers.

2 TREMOR MEASUREMENT

The tremor measurement was executed on four

tremor patients as a preliminary experiment using an

accelerometer and the data was analyzed by FFT

(Fig.1). They have 2-4Kg weights in order to cause

their tremor.

Figure 1: Tremor measuring system.

The obtained tremor graph is clearly

distinguished in two groups, namely Parkinson

tremor and Physiological tremor (Fig.2).

Figure 2: Parkinson tremor and Physiological tremor.

Measured tremor graph is analyzed by FFT

(Fig.3) and we have found that the main frequency

of Parkinson tremor is about 4Hz and the

physiological tremor is about 8Hz in arms. The

severity of tremor is classified in five levels from 0

(no tremor) to 4 (most severe) subjectively by a

tremor specialist. The two tremor groups are plotted

in one graph by the tremor severity level (Fig.4).

They can also be divided in two groups clearly and

Parkinson tremor has tendency of negative

correlation between the severity and the main

frequency.

164

Fukumoto I..

A Study for Automatic Diagnosing System of Parkinson Disease - A Systematic Analysis of Parkinsonian Tremor by Accelerometer.

DOI: 10.5220/0005016301640168

In Proceedings of the 11th International Conference on Informatics in Control, Automation and Robotics (ICINCO-2014), pages 164-168

ISBN: 978-989-758-039-0

Copyright

c

2014 SCITEPRESS (Science and Technology Publications, Lda.)

Figure 3: FFT analysis of tremor.

Figure 4: Correlation between tremor severity and main

frequency.

Six Parkinson patients in L-Dopa treatment are

examined using the measuring system. The tremor

power in the accelerometers’ output (mV) is plotted

in the same graph (Fig.5).

Figure 5: Tremor change by L-Dopa treatment.

Almost all patients except one show the

increased main frequency and the decreased tremor

power. The one patient has On-Off phenomenon

that means the failure in pharmacological treatment.

3 MODEL & SIMULATION

There are two hypotheses on the cause of Parkinson

tremor namely the central oscillation theory and the

peripheral feedback theory. The former thinks that

there are oscillators in bran stem that may activate

skeletal muscle to elicit tremor at certain determined

tremor frequencies. The latter insists that the tremor

may be caused by the feedback loop between the

skeletal muscle and the motor neuron in bone

marrow. The obtained data from our experiment

show that the main frequency can be changed by

medical treatment, which means the peripheral

feedback theory is more likely true.

But it is almost impossible to be certified

because human brain is too complicated and very

difficult to be analyzed anatomically. Computer

simulation on the mathematical model is another

way to prove the hypotheses.

Auto-regression analysis (AR) is executed in

order to get the number of independent parameters

on the Parkinson patients’ tremor (Fig.6). It shows

that eight independent parameters are enough to

describe the tremor producing system.

Figure 6: AR model analysis of Parkinson Tremor.

A mechanical oscillating model is constructed

for the human forearm tremor (Fig.7).

Figure 7: Forearm oscillation model.

The muscle spindle is also made using central

modifiable intra spindle muscle (Fig.8).

AStudyforAutomaticDiagnosingSystemofParkinsonDisease-ASystematicAnalysisofParkinsonianTremorby

Accelerometer

165

Figure 8: Intra muscle spindle model.

A mathematical formulation is described in

simple equations.

The rotation of the forearm due to the torque N(t)

is described by defining the inertia of the forearm J,

the elastic coefficient of the biceps Km, viscosity

coefficient Dm, the deviation of Q from the initial

angle Q

0

, θ, so that Q=Q

0

+ θgiving

Let the viscosity coefficient of the intrafusal

fiber be Ds and the mass of the intrafusal fiber be

negligible.

Using the elastic coefficient of the intrafusal

muscle fiber Ks, a non-dimensional effectively

coefficient appears including simulating effect by

gamma neuron F

gk,

the unit of which is the number

needed to stretch the intrafusal muscle fiber by half

its original length (c.a. 6mm).

F

gk

may

be thought to indicate the level of the

intrafusal muscle.

Computer simulation using the mathematical

model is executed by varying the F

gk

value (Fig.9).

Figure 9: A Simulation result varying F

gk.

Most remarkable result of the simulation may be

the sudden change of the main frequency by F

gk

(Fig.19).

Figure 10: Parkinson tremor suddenly changes to

Physiological tremor by F

gk

increasing.

The graph shows that the main frequency can

take only two state namely the Parkinson tremor

(4Hz) and Physiological tremor (8Hz) and it cannot

take the middle value. The fact corresponds well to

the clinical observation and our measured data.

Main frequency obtained from 50 Parkinson patients

are plotted against their The Hoehn and Yahr scale

(

□) (Fig.11).

Main frequencies calculated from corresponding

F

gk

value are also plotted in the graph (○). The two

plots vary simultaneously and the fact may support

the based hypothesis of peripheral feedback

oscillation.

ICINCO2014-11thInternationalConferenceonInformaticsinControl,AutomationandRobotics

166

Figure 11: Yahr scale vs. Main frequency of Parkinson

patients.

4 BIOFEEDBACK TRINING

Medical treatment of Parkinson’s disease has several

shortcomings such as Wearing-Off or On-Off

phenomena as well as side effects of involuntary

movement. Biofeedback training is new non-

invasive treatment for psychological diseases. We

have applied the tremor measurement method to the

biofeedback training of Parkinson tremor.

Figure 12: Biofeedback system of Parkinson tremor.

Figure 13: Tremor biofeedback with accelerometer.

Biofeedback needs biological signal sensor and

feedback unit of the signal to patients. The feedback

signal is tremor curve in visual biofeedback and

modified pure sinusoidal tone in sound biofeedback.

The visual biofeedback is executed on three healthy

student (♀1♂2).

Figure 14: Visual biofeedback.

The subjects listen to the pure tone of which

loudness changes according to the tremor frequency

and they try to decrease the tone amplitude.

The tremor power (amplitude) decreases during and

after the biofeedback training (Fig.14). The main

frequency of tremor increases by the visual

biofeedback (Fig.15).

Figure 15: Tremor change by visual biofeedback.

Sound biofeedback uses pure sinusoidal tone

modified by tremor power as the feedback signal

instead of the tremor curve. The sound biofeedback

is executed on ten Parkinson patients ( ♂ 2 ♀ 8,

74.9±5.1years old).

Figure 16: Tremor change by sound biofeedback.

The training effect shows the decrease of tremor

power (p<0.05) and the increase of main frequency

(p<0.05) (Fig.16).

AStudyforAutomaticDiagnosingSystemofParkinsonDisease-ASystematicAnalysisofParkinsonianTremorby

Accelerometer

167

Both L-Dopa medical treatment and the two

biofeedback trainings show the same effect but the

medical is most effective (Fig.17).

Figure17: Comparison of three methods for the tremor

improvement.

5 CONCLUSIONS

Parkinson Disease can be measured and diagnosed

objectively with 3 D accelerometers. A

mathematical model based on the peripheral

feedback theory is tested by computer simulation

with good coincident with the clinical data. We

have found the main frequency of Parkinsonian

tremor is about 4Hz in arms. Using our method the

tremor improvement by biofeedback training is

objectively proved without any arbitral evaluation.

The result that the main frequency of tremor

increases and the tremor power decreases according

to the treatments corresponds well to clinical fact.

ACKNOWLEDGEMENTS

I would like to extend my sincere thanks to my

former students; Yoshinobu MATSUMOTO, Hisashi

UTIYAMA, Kiyoshi OKADA, Kenshi SUZUKI and

Takahiro YOSHII who have executed experiments

earnestly, as well as DR.Masato TAMURA &

Dr.Yasuhiro KAWASE who have assisted clinical

measurement in their hospitals

.

REFERENCES

Fukumoto, I., 1986. "Computer simulation of

Parkinsonian Tremor", J. Biomed. Eng., Vol.8, pp50-

55.

Fukumoto, I., 1994. "Computer simulations of human

tremors", Abstracts of the World Congress on Medical

Physics and Biomedical Engineering Part2 [653], Rio

de Janeiro, Brazil.8

Matsumoto, Y., Shindo, K., Sakakibara, M., Tamura, M.,

Fukumoto, I., 1996. "Study of the 8Hz Tremor

Frequency Related to Dopamine Level in Parkinsonian

Patients", IEEE.EMBS96, Final Program & Abstract

Book, Amsterdam, pp122

Utiyama, H., Fukumoto, I., 1999. “The basic studies of

biofeedback treatment systems for Parkinson's tremor,

dementia and visual accommodation disorders.

Proceedings of AAPB 13th Anniversary Annual

Meeting, P231, Vancouver

Okada, K., Hando, S., Teranishi, M., Matsumoto, Y.,

Fukumoto, I., 2001. “Analysis of pathological tremors

using the autoregression model", Frontiers

Med.Biol.Engng, Vol.11, No.3, pp221-235

Okada, K., Fukumoto, I., 2002. "A basic analysis of

pathological tremors using the autoregression model",

Proc. of Int. Cong. on Biological and Medical

Engineering, D3VA-1055, Singapore

Matsumoto, Y., Yoshii, T., Tamura, M., Fukumoto, I.,

2005. ”The change of diseased tumor acceleration

parameters under the photic feedback stimulation”,

Transactions of the Japanese Society for Medical and

Biological Engineering, Tsukuba, pp648, Apr25-27

Fukumoto, I., 2006. ”A Unified Model of the Alzheimer-

Dementia and Parkinsonian-Disease Based on the

Somatic Stimulation Effects”, WC2006 program book,

pp125-128, Seoul

Matsumoto, Y., 2009. Ichiro Fukumoto: ”Development of

pathological evaluating system of tremor disease by a

tablet PC and accelerometers”, EC2009, 07/09.14,

Munchen

ICINCO2014-11thInternationalConferenceonInformaticsinControl,AutomationandRobotics

168