In: Proceedings of the 19th International Conference – IEEE/EMBS Oct. 30. – Nov. 2., 1997 Chicago, Il. USA pp. 1760-1763.
ANALYSIS OF MOVEMENT PATTERNS AIDS THE EARLY DETECTION OF PARKINSON’S DISEASE
Ákos Jobbágy*1, Hans Furnée*2, Péter Harcos*3, Miklós Tárczy*4, Ivan Krekule*5, László Komjáthi*1

 

*1: Department of Measurement and Instrument Engineering, Technical University of Budapest
Mûegyetem rkp. 9. H-1111 Budapest, Hungary e-mail: jobbagy@mmt.bme.hu
*2: Delft University of Technology, Lorentzweg 1, 2600 GA Delft, The Netherlands
*3: Szt.Imre Hospital, Tétényi út 12/16. H-1115 Budapest, Hungary
*4: Semmelweis University of Medicine, Üllõi út 26. H-1085 Budapest, Hungary
*5: Czech Academy of Sciences, 14220 Prague 4, Videnska 1083, Czech Republic

 

Abstract: The movement patterns are characteristic for a person. The changes in his/her movement patterns indicate changes in his/her state. Several diseases result in well distinguishable deviation from normal movement patterns. The deviation starts with subtle changes, our aim is to detect these changes and thus aid the early diagnosis of neurological diseases. We have been using a passive marker based motion analyzer (PRIMAS) to record the movement patterns defined by neurologists. Healthy subjects and Parkinsonian patients participated in the tests. We suggest a measurement set-up and evaluation algorithms that fit to the movement patterns to be analyzed.

 

 

INTRODUCTION
 

The therapy of a patient with a neurological disease is the more effective the earlier it is started. Neurologists observed characteristic changes in the movement patterns of their patients a long time ago. The results of several tests were published, when the patients were known to be Parkinsonians and the level of movement distortion had to be qualified. Good examples are given in [1], [2] , [3] and [4]. In the early, preclinical phase, the subtle changes cannot be detected by visual inspection. With the help of a motion analyzer the very early signs of movement distortions can be revealed. Our aim is to define a screening test for neurological diseases, especially for Parkinson’s disease. Contrary to the conventional tests, we aim at specifying a set of movement patterns that help in establishing the diagnosis for patients whose condition is not previously known.

The movement patterns were defined by neurologists. These are hand- and finger movements, it takes only 15 minutes per person to record them. We applied a passive marker-based motion analyzer. Passive markers can easily be attached to anatomical landmark points of patients. These markers influence the movements only negligibly. Fig. 1 shows the markers attached to the middle phalanges of a person.

We have developed evaluation programs for all the movement patterns. These result in a single parameter characterizing the level of deviation from normal. This parameter can help in the assessment of the treatment as well.

 

 

Figure 1 Passive markers attached to the middle phalanges of a person
 
 
RECORDING MOVEMENT PATTERNS
 

The Precision Image-based Motion Analysis System (PRIMAS) has been applied for recording movement patterns. This system is based on the video/digital conversion method developed at Delft University of Technology. PRIMAS is able to determine the 3D position of markers in real-time at a 100 frame/s rate [5], [6]. Assuming average size markers the resolution of the system both horizontally and vertically is 1:6000 of the corresponding side of the field of view (FOV). For our recordings of hand- and finger movements this assures an 0.1 mm resolution.

During the recording the subjects remain seated on a chair. There is a table in front of the chair except during the twiddling movement. There are marked points on the table that ensure the reproducibility of the recordings.

Complex movement patterns have to be specified, as simple movements can be performed by Parkinsonians with relatively small deviation from the control group. The movement patterns aim at revealing the manifestations of muscular rigidity and bradykinesia.

 

The persons we tested performed the following movement patterns:

Tapping: Patients put their hands on the table, 10 mm diameter markers are attached to the middle phalanges of their fingers. Patients lift 8 of their fingers (thumbs remain on the table) and then hit the table with their fingertips in the following order: little-, ring-, middle- and index finger. This mimics the piano-playing movement.

Twiddling: Patients twiddle their hands in front of their trunks. Forearms are nearly horizontal. Markers with 25 mm diameter are attached to their forearms, approximately 20 cm far from their carpal bones.

Pinching and circling: This is performed in 6 phases. The 4 single hand movements are: pinching with the right hand, pinching with the left hand, circling with the right hand, circling with the left hand. The 2 parallel movements are: pinching with the right hand while circling with the left hand and pinching with the left hand while circling with the right hand. Markers with 10 mm diameter are attached to the middle phalanges of the index fingers.

Pinching movement: The forearm is in vertical position, little-, ring- and middle fingers touch the palm. The patient moves his/her index finger: touches the thumb, then lifts the index finger and then touches the thumb again.

Circling movement: The forearm is in horizontal position, the index finger is stretched. The forearm circles around the superficies of an imaginary cone, the elbow is supposed to remain in nearly the same position.

Each movement pattern was performed repeatedly for 8 seconds, parts of the recordings are shown in Figs. 2-7. In these figures the time functions of the vertical coordinates of the marker positions are shown. The figures show some movement patterns of three persons: a young- and a senior healthy subject and a Parkinsonian patient. All persons were asked to perform each movement pattern as fast as they could.

Figure 2 Pinching and circling, young healthy subject
 
Figure 3 Pinching and circling, senior healthy subject
 
Figure 4 Pinching (right, affected hand, solid line) and circling (left hand, dashed line), Parkinsonian patient
 
Figure 5 Pinching (left hand, dashed line) and circling (right, affected hand, solid line) Parkinsonian patient
In case of Parkinsonian patients the pacing can help in movement coordination thus we did not use paced tests for the early detection.

Though a small fluctuation in the phase shift can be observed between the movements of the two hands in Figs. 2 and 3, these are well coordinated movements. In contrast to them, the movement is not well coordinated in Figs. 4 and 5.

When the Parkinsonian patient performed the pinching with his right (affected) hand, this movement was irregular. When the pinching was performed with his left hand, an involuntary circling of this hand appeared. The involuntary circling is synchronized to the circling of the other hand.

Figs. 6-7 show twiddling movements, Fig. 6 is of a young healthy subject, Fig. 7 is of a Parkinsonian patient.

 

Figure 6 Young patient, twiddling

 

Figure 7 Parkinsonian patient, twiddling. Affected right hand shown with solid line

 

 

EVALUATION OF RECORDED DATA

 

Though the human image processing ability is excellent, the processing of the recorded time functions by human visual evaluation would not yield accurate results to aid early diagnosis. Notice the varying amplitude and delay between the functions that would be very difficult to express numerically using manual evaluation. The calculation of the exact repetition rate and its stationarity would be equally difficult. The analysis and evaluation methods were discussed by experts from different fields and tested using the basic evaluation programs and the whole recorded data base (time functions of all markers).

We can easily evaluate Figs. 2-4: the first two are very similar to each other while the third is markedly different from them. The regularity of the third recording is much smaller than that of the other two. The figures also show that in case of the Parkinsonian patient the frequency of circling was much slower (about one fifth) and the amplitude of circling was smaller (about one half) than for healthy subjects. The exact definition of the evaluation algorithms is necessary for the quantification and for the automated processing. This makes information reduction possible: the selection of the important parameters that describe movements accurately enough for the goals to be achieved. Once the parameters are selected, classes can be defined, such as normal and different pathological behavior of the examined subject. A class allows for a range of parameters, a certain overlap of classes cannot be avoided, the realistic aim is to minimize such overlaps.

The following parameters have been found to be relevant in the case of finger-, hand- and arm movements: frequency, amplitude, smoothness, symmetry, effect of the movement of the other hand. The feature extraction proved to be the most difficult when the two hands performed different movements. Nevertheless, these movements have the strongest diagnostic power.

The detailed description of the evaluation of the tapping movement is given in [7]. In the following we analyze the pinching and circling as well as the twiddling motion in more detail. The pinching and circling movement has been characterized by 4 parameters, pici1,..,4. The first expresses the frequency stability of each hand and finger movement. 8 movement patterns belong to the six phases of this test. The FFT is computed for each pattern and the energy content around the dominant frequency is compared to the total energy content:

Persons can achieve higher frequencies when moving only one of their hands. The second parameter shows the slowing down of the frequency of two-hand movements compared to single hand movements. The dominant frequencies of the 4 one-hand movements are averaged and compared to the average of the 2 two-hands movements:

 

The third parameter shows the asymmetry of the two hands: the difference of the maximum dominant frequencies reached by the left and right hands:

 

 

Higher frequencies can be achieved if the amplitude of pinching is smaller. The fourth parameter takes it into account by multiplying the frequency (fp) and amplitude (Ap) of pinching:

pici4 = Apfp

 

The twiddling movement is characterized by two parameters, twi1 and twi2. The first parameter stands for the symmetry of twiddling. For Parkinsonian patients the hand not affected circles around the affected one thus the amplitudes (A) of the two hand movements are different, cf. Fig. 7.

 

 

The second parameter is the dominant frequency of twiddling. The dominant frequency is compared to 4 Hz:

 
 

If the dominant frequency is higher than 4 Hz, twi2 = 0.

 

To characterize the state of a subject a single value is computed based on all the parameters describing his/her movement patterns. At present there is no weighting, all parameters have equal effect on the final value. The global qualification parameter is normalized to the 0 ... 1 range. According to our experience, the tapping movement and the slowing down during pinching and circling (pici2) have the highest diagnostic power.

 
 

      global qualification parameter
      young subjects
      senior subjects
    Parkinsonian patients
    minimum
      0.1637
      0.1899
      0.4142
    maximum
      0.3446
      0.3580
      0.6012
    average
      0.2369
      0.2568
      0.4945
 
Table I The global qualification of the tested patients

 

 

We recorded the above movements of 12 young (20..25 year old)- 10 senior (50...65 year old) healthy subjects and 3 Parkinsonian patients (50...65 year old). The results are summarized in Table I.
 
 

CONCLUSIONS

The recordings demonstrate that subtle deviations can be detected in the movement patterns of subjects. The definition of the gold standard requires the accurate diagnosing of the patients by neurologists using conventional methods. This is especially difficult for those being in the early phase of Parkinson’s disease. Sophisticated and expensive measurements (PET or SPECT) can objectively reveal the suspicion of the disease in the preclinical phase.

The test we suggest can be done in 15 minutes thus it may be used as a screening test. An effective screening test would require the standardization of the measurement set-up, the movement patterns and the evaluation algorithms. An international cooperation has been formed in order to define and validate a screening test, involving Czech, Dutch, French, Hungarian and Slovenian experts. Neurologists, psychologists, electrical engineers and human movement scientists are among the partners. The data evaluation shows that for the screening test a much simpler motion analyzer would be satisfactory. A simple and cheap 2D analyzer is being developed based on a commercial video camera and a personal computer.

 
 
REFERENCES
[1] van Emmerik, R.E.A., Wagenaar, R.C.(1995) Tremor and symmetry properties in bimanual coordination in Parkinson’s disease. In: Bardy, B.G., Bootsma, R.J. & Guiard, Y. (eds.) Studies in Perception and Action. Lawrence Erlbaum, pp. 61-64.

[2] Muir, S.R., Jones, R.D., Andreae J.H., Donaldson, I.M. (1995) Measurement and Analysis of Single and Multiple Finger Tapping in Normal and Parkinsonian Subjects. Parkinsonism & Related Disorders, Vol. I. No. 2. pp 89-96.

[3] Johnels, B., et al.,(1987) Measuring motor function in Parkinson’s disease. In Motor Disturbances I, eds. R. Benecke, B. Conrad, C.D. Marsden. Academic Press, 1987. pp 131-144.

[4] Furnée, E.H., Jobbágy, Á.(1993) Precision 3-D Motion Analysis System for Real-time Applications. Microprocessors and Microsystems, Vol. 17. Nr.4. May 1993, pp. 223-231.

[5] Valls, M. et al.(1990) Age-related slowing and fragmentation of a complex movement quantified by optoelectronic kinesiology. Clinical Rehabilitation, 4, pp. 111-122.

[6] Jobbágy, Á. et al. (1995) Biomedical Applications of a Precision Motion Analysis System. Proc. of the 7th Int. IMEKO TC-13 Conf. Model Based Biomeasurements, pp. 401-403.

[7] Jobbágy, Á. et al. (1997) Automatic Movement Evaluation for the Early Detection of Parkinson’s Disease. IEEE EMBS Magazine, accepted paper