Analysis of neurological disorders based on digital processing of speech and handwritten text

The paper deals with the methods of non-invasive analysis of neurological disorders, focusing on speech signal processing and processing of handwritten text. The paper describes the whole procedure of the automated analysis of the disorder while the greatest attention is paid to a parameterization. In the case of speech signal analysis, the state-of-the-art features evaluating a presence of hoarseness, breathiness and hypernasality are mentioned. Nonlinear dynamic parameters and parameters derived from the empirical mode decomposition (EMD) are compared. Based on the tests, from the point of description of a noise component of signal, the best results were obtained using the approximation entropy, the largest Lyapunov exponent and parameters based on Teager-Kaiser energy operator, which is calculated from the first intrinsic mode function (IMF). In the case of handwritten text analysis, the most used exercises describing a tremor and movement dynamics are mentioned. The new approaches of hand movement analysis at a time when the pen tip does not touch the paper have been also proposed. Finally the paper discusses different applications of speech signal and handwriting text parameterization.