Název:	Heterogeneous Medical Data Processing
Autoři:	Prokop, Tomáš
Datum vydání:	2017
Nakladatel:	Západočeská univerzita v Plzni
Typ dokumentu:	report zpráva
URI:	http://www.kiv.zcu.cz/cz/vyzkum/publikace/technicke-zpravy/ http://hdl.handle.net/11025/35988
Klíčová slova:	elektroencefalografie;zpracování signálu;Hilbert Huangova transformace
Klíčová slova v dalším jazyce:	electroencephalography;signal processing;Hilbert-Huang Transform
Abstrakt v dalším jazyce:	Electroencephalography (EEG) is a popular technique used for example in diagnostics of diseases, sleep monitoring and neurorehabilitations. Due to increasing mobility and decreasing price of EEG measuring devices EEG and ERP (event related potential) techniques have become more widespread also in assistive technologies and brain-jogging. As the result the amount of EEG data has been increasing and research into EEG signal processing and classification has become again more necessary. This thesis focuses on the state-of-the-art related to the methods of EEG/ERP signal processing and classification, but follows a standard processing workflow starting from signal acquisition and preprocessing to feature extraction and classification. The commonly used time-frequency domain methods (Wavelet transform and Matching pursuit) that are suitable for feature extraction are described together with the Hilbert-Huang Transform (HHT). HHT uses a new approach of multi-channel signal decomposition called the Multivariate Empirical Mode Decomposition. The described classification methods are divided into two groups. Linear classifiers are represented by the Linear Discriminant Analysis and Support Vector Machines. The second group, neural networks, focuses on the Multi-Layer Perceptron and a set of classification algorithms called deep learning neural networks. These are composed of many layers of neurons while the Multi-Layer Perceptron typically contained only two layers because of the backpropagation problem. Some of the deep learning algorithms have been reported to beat state-of-the-art approaches in many applications and that is why further research in the EEG domain seems to be beneficial.
Práva:	© Západočeská univerzita v Plzni
Vyskytuje se v kolekcích:	Zprávy / Reports (KIV)

Všechny záznamy v DSpace jsou chráněny autorskými právy, všechna práva vyhrazena.