Mathematical approach to tremor recognition in extrapyramidal disease

Tremor is rhythmic involuntary oscillatory movement of body parts. There have been attempts to separate tremors due to different extrapyramidal diseases with objective techniques by measuring the acceleration of hands and applying the features of conventional spectral description, peak value and peak frequency of power spectral density (PSD). However, no clear-cut separation has been found with these methods; none of the parameters helps to decide reliably which kind of pathological tremor is present. The task is then twofold: to exploit other PSD features and to apply descriptions extracting new information from existing data. This article shows that the features of this process, based on the application of cumulants and polyspectra, represent different types of tremor and may stand as the basis for recognition between them. The results of numerical experiments are presented and discussed.