Algebraic estimator of Parkinson’s tremor frequency from biased and noisy sinusoidal signals

Abstract

Tremor is an uncontrolled trembling movement or shakes, which are defined as an involuntary, rhythmic oscillatory movement of the body. The dominant features of Parkinsonism are the motor task and its frequency. This paper presents studies on the tremor parameter identification to be used for obtaining the frequency as a dynamical feature of the tremor. The method is based on the analysis of time-varying signals for identification of the tremor’s frequency from unknown noisy harmonic signals with an offset, using time-varying unstable filters and low-pass Butterworth filters. This approach uses an algebraic derivative method, in the frequency domain, to obtain the main frequency of tremors in the time domain. The first frequency mode of the tremor is one of the main characteristics to represent the low vibrational dynamics of Parkinson’s tremor. The proposed frequency estimation is performed in less than a period of the slower component of the measured signal. Real tremor signals were used to experimentally validate the proposed method and the algorithm proved to be fast and robust to high-frequency noises tracking the time variation of the tremor accurately.