A recording and analysis system for human tremor Mehmet Engin * Ege University, Faculty of Engineering, Electrical and Electronics Engineering Department, 35100 Bornova, _ I zmir, Turkey Received 4 May 2005; accepted 30 May 2006 Available online 7 June 2006 Abstract There are various diagnosis and treatment monitoring are available for many neuro-muscular diseases. This paper is concerned with the tremor analysis for the purpose of recognition. A recording and analysis system for human’s tremor are developed. The analysis was performed based on frequency and amplitude parameters. The fast Fourier transform (FFT) and higher-order spectra were used to extract frequency parameters (e.g., main peak frequency). In order to diag- nose subjects’ condition, classification was implemented by statistical significant tests (t-test). Ó 2006 Elsevier Ltd. All rights reserved. Keywords: Tremor analysis; FFT; Higher-order statistics; Data acquisition; Statistical test 1. Introduction Tremor is phenomenon observable at either healthy population or at the people with pathologi- cal changes. Tremor time series are considered as stochastic signals [1]. Because the mechanism of gen- erating them is so complex and exposed to so many uncontrollable influences that the mathematical equations describing them ought to contain random quantities. Following classes of tremor are the most impor- tant: the physiological, essential, and parkinsonian [1]. The time signal from person with physiological tremor could be often considered as the realization of linear stochastic process. On the other hand, the time series coming from parkinsonian patient can be identified as signals of non-linear chaotic nature. The essential tremor occupies the position in the middle between the two extremes, the physiological and parkinsonian tremor. Among the pathological cases, parkinsonian and essential tremor are the most often observed types. Parkinson’s disease is a growing problem with 120–180 victims in every 100,000 people. Although most patients over the 40 years old are parkinsonian, the disease can also appear in younger subjects [2]. While investigators have discovered many of the brain’s mechanism, the causes of the disease are still unknown, and the clinical distinction can be difficult. The electrophysiological analysis of human tre- mor has a long tradition. In 60s, different techniques were used to measure the amplitude and frequency of tremor signals. Nowadays, with the fast develop- ment of digital computers, one began to sample the tremor records to analyze the data off-line mainly by the new developed methods of spectral and cross-spectral analysis of the stochastic process [3]. 0263-2241/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.measurement.2006.05.015 * Tel./fax: +90 232 388 60 24. E-mail address: mengin@bornova.ege.edu.tr Measurement 40 (2007) 288–293 www.elsevier.com/locate/measurement