Funkční konektivita zdrojů EEG pro časově vázané události

Abstract

Studie zabyvajc se vyzkumem principu vedom jsou uzce svazany s elektrickou aktivitu mozku a jejm rozlozenm na povrchu hlavy. Tuto aktivitu v okamzicch casove vazanych udalost (ERP) zaznamenavame pomoc elektroencefalografu (EEG). Hlavnm clem techto studi byva detekce propojen jednotlivych center mozku v ERP. Vyhodnocovan e segmenty ERP jsou ve sve delce omezeny z duvodu stochasticke povahy EEG zaznamu a jsou obtzne vyhodnocovany. Metodou vystizne zobrazujc konektivitu mezi aktivnmi centry v mozku je funkcn konektivita lokalizovanych zdroju EEG. Zpresnen vyslednych hodnot lze zskat za pouzit individualnch modelu hlavy subjektu. Clem diplomove prace bylo vytvorit navrh metodiky hodnocen, analyzy a zpracovan ERP 256 kanaloveho EEG signalu se zamerenm na lokalizaci zdroju EEG signalu a jejich funkcn konektivitu ve frekvencnm pasmu gama s pouzitm individualnch modelu hlavy subjektu. Pro implementaci navrzene metodiky bylo urceno prostred MATLAB s Fieldtrip toolboxem. Metodika byla vytvorena za ucelem vypomoci Narodnmu ustavu dusevnho zdrav (NUDZ) ve forme softwaru ke studii " Role gamma synchronizace na vzniku vedome zkusenosti\ (Game). Zpracovavana EEG data, vyuzita k potvrzen funkce metodiky, byla namerena v NUDZ na 13 dobrovolncch. Dobrovolnci byli muzi i zeny ve veku mezi 18 - 60 roky. Pocet zaznamenan ych ERP v zaznamu se pohybuje v rozmez 18 - 100 trialu. Pro 3 dobrovolnky z 13 zmnenych byly vytvoreny snmky z magneticke rezonance (MRI), pouzitelne k vytvoren individualnch modelu hlavy. Zskana data celych zaznamu s 256 kanaly byla segmentovana do casovych interval u obsahujcch ERP. Tato vybrana data byla ltrovana s pomoc FIR tru za ucelem zskan cistych hodnot EEG signalu ve frekvencnm pasmu gama. Po aplikaci ltrace byly z ltrovanych segmentovanych dat vybrana data obsahujc pouze zajmovou oblast ERP. Z vysledneho poctu 483 segmentu byly na zaklade zdrojove literatury vybrany tri casova okna o delce 100 ms denujc pozice v gama synchronizacnm signalu. Vzorky byly podrobeny studentovu statistickemu t-testu za ucelem zjisten statistickeho rozdlu aktivac ve frekvencnm pasmu gama pri vstupu podnetu do vedom. Na zaklade vysledku statistickeho zpracovan byly potvrzeny pozice pro vyuzit lokalizace zdroju a vypoctu funkcnch konektivit. MRI snmky zskane v ramci studie Game v NUDZ od prvnch 3 dobrovolnku byly zpracovany v programovacm prostred MATLAB za vyuzit eldtrip knihovny, a byly z nich vytvoreny individualn modely hlav. Tyto modely byly vyuzity v lokalizaci zdroju EEG signalu na individualn mrzce pozic upravene stejnym modelem hlavy v casovych intervalech denovanych analyzou EEG. Vysledne hodnoty zdroju byly parcelovany za pomoc vybraneho atlasu lokalit a byly vytvoreny konektivitn matice koherenc, zobrazujc propojen oblast v okamzicch zajmu.

Research studies of consciousness principles are closely linked to the electrical activity of the brain and its distribution on the surface of the head. We record this activity at times of time - bound events (ERP) with electroencephalography (EEG). The main objective of these studies is to detect interconnection of individual brain centers in ERP. Evaluated ERP segments are limited in their length because of the stochastic nature of the EEG record and are dicult to evaluate. A method that displays the connectivity between the active centers in the brain is the functional connectivity of localized EEG sources. The renement of the resulting values can be obtained using individual head unit models. The aim of this diploma thesis is to design a methodology for evaluation, analysis and processing of ERP in 256 channel EEG signal with focus on localization of EEG signal sources and their functional connectivity using individual heads of subjects. The MATLAB environment with the eldtrip toolbox is selected for design of the methodology. The methodology was developed to help the National Institute of Mental Health (NUDZ) in the form of software for Game study. The processed EEG data used to conrm the function of the methodology was measured in the NUDZ for 13 volunteers. The volunteers were men and women between the ages of 18 - 60. The number of recorded ERPs in the record ranges from 18 to 100 trials. For 3 of the 13 volunteers, magnetic resonance imaging (MRI) images were used to create individual head patterns. The full data of the entire records with 256 channels was segmented into time intervals containing ERP. These selected data were ltered using FIR lters to obtain pure EEG signals in the gamma frequency band. After ltering, data that contains only the ERP interest area was selected from the ltered segmented data. From the resulting number of 483 segments, three time windows of 100 ms were selected based on the source literature, dening positions in the gamma synchronization signal. The samples were subjected to a student's statistical t-test in order to determine the statistical dierence between activations in the gamma frequency band when the stimulus enters the consciousness. As a result of the statistical processing, the positions for utilization of resource localization and the calculation of functional connectivity were conrmed. The MRI images obtained in the NUDZ Game trial from the rst 3 volunteers were processed in the MATLAB programming environment using the eldtrip library and created individual headmodels. These models were used to locate EEG signal sources on individual grid positions treated with the same headmodel at time intervals dened by EEG analysis. The resulting resource values were parcels using the selected atlas of the sites, and connectivity matrices were created, showing the interconnection of the areas in moments of interest.