Comparison of methods for tremor frequency analysis for patients with Parkinson's disease

Abstract

Tremor is described as involuntary rhythmic oscillations of one or more body parts. It is a symptom of Parkinson's disease (PD). The severity of tremor is based on its frequency. Using acceleration sensors, one can detect tremor of the limbs or other body parts. Data from sensors can be processed using spectral analysis. The most common methods for the investigation of tremor are Fast Fourier Transformation (FFT), Short Time Fourier Transform (STFT) and power spectral density analysis (PSD). In this paper we investigate these methods together with peak detection and pattern recognition methods. We compare the various approaches with each other with respect to frequency. A visual frequency analysis using an optical tracking system is used as a reference. The experiments were performed with a measuring glove with integrated acceleration sensors on the middle finger and thumb joint. We examined the accuracy of the various methods for the analysis of tremor in PD patients.