Quantifying Bradykinesia in neurological patients

A.C. Schouten¹, A. Munts², H.E.J. Veeger¹, F.C.T. van der Helm¹, J.J. van Hilten²

¹Delft University of Technology, Department of Mechanical Engineering, The Netherlands
²Leiden University Medical Center, Department of Neurology, The Netherlands

Introduction

Patients with neurological disorders, like Parkinson’s disease (PD), often show bradykinesia. Bradykinesia is demonstrated by slowness and irregularities during fast voluntary movements. Physicians use this sign as an indication for lack of neuromuscular control. Normally bradykinesia is visually scored by a physician. Patients are requested to perform a .nger task, i.e. open and close the thumb and index finger as fast as possible, while the performance is scored, grading from 0 (normal) to 4 (can barely perform). This visual score is subjective, physician dependent and has a small resolution. Goal of the project is to develop a method that objectively and accurately scores bradykinesia.

Method

Subjects (PD, n=14; controls, n=36) were requested to perform a finger task for 15 seconds under a high speed digital camera (60 fps). The subjects were instructed to close and open the thumb and index finger as wide and as fast as possible. Markers of colored tape (Leukotape®) were attached to the top of the thumb (green) and index finger (blue). With special software (EthoVision^®, Noldus Information Technology bv, The Netherlands) the center of the markers was located with time. As the distance of the hand with respect to the camera was fixed the distance between the fingers could be calculated. The data were further processed using Matlab and two parameters were selected to represent the slowness of the movement: median frequency (MF) and mean absolute acceleration (MA).

Results and discussion

Patients with PD had a significantly lower value for both MF (controls: 3.56 Hz (±0.94); PD: 2.25 Hz (±0.90)) and MA (controls: 21.3 m/s2; (±7.36); PD: 9.76 m/s2 (±5.05)). Statistical analysis revealed that both parameters depended on age and sex, which should be taken into account; handedness was not significant. With a larger dataset we will focus on the most differentiating parameter to describe movement slowness. MA is the primary candidate as acceleration depends on both the distance and speed of the movements, and such is less sensitive to task variations (wide vs. fast). Finally, it is concluded that movement slowness can be objectively quanti.ed, showing the relevance of the method..

Paper presented at Measuring Behavior 2005 , 5^th International Conference on Methods and Techniques in Behavioral Research, 30 August - 2 September 2005, Wageningen, The Netherlands.

© 2005 Noldus Information Technology bv