Citas bibligráficas
Salazar, I., (2018). Evaluación de técnicas del problema inverso para estudio del número de señales de electroencefalograma para mecanismos cognitivos [Tesis, Universidad Nacional de Ingeniería]. http://hdl.handle.net/20.500.14076/18141
Salazar, I., Evaluación de técnicas del problema inverso para estudio del número de señales de electroencefalograma para mecanismos cognitivos [Tesis]. : Universidad Nacional de Ingeniería; 2018. http://hdl.handle.net/20.500.14076/18141
@mastersthesis{renati/710238,
title = "Evaluación de técnicas del problema inverso para estudio del número de señales de electroencefalograma para mecanismos cognitivos",
author = "Salazar Reque, Itamar Franco",
publisher = "Universidad Nacional de Ingeniería",
year = "2018"
}
Electroencephalography is a non-invasive method for brain biological signals monitoring. This method has several uses in both research and clinical applications such as mental disabilities. To analyze them, models based on signal averaging are used. Such methods are few realistic and they don’t allow researchers to appreciate the interaction of the different cortical areas and their influence on the captured signals, thus reducing the capacity of analysis. Some models, as Dynamic Causal Modelling (DCM), aim to analyze EEG signals in a more realistic way, considering the interaction of neurons and their different connectivities, as well as to infer synaptic properties associated with behavior. To apply these models, it is required to know the localization of neuronal sources underlying EEG signals. To do that, several techniques have been developed to estimate the localization of these sources using only EEG signals taken from the scalp, a problem which is commonly known as the “EEG Inverse Problem”. These techniques commonly use many electrodes (>64). On the other hand, it is common, for clinical studies, to analyze EEG data with lesser quantity of electrodes (<40). Thus, the use of techniques to solve the inverse problem with fewer number of electrodes is very important. In this dissertation, performance of several techniques (MNE, WMNE, LORETA, FOCUS, MFOCUSS y MSP) to solve the EEG inverse problem was analyzed. Performance was measured in terms of error localization of the aforementioned sources for synthetic EEG signals with different levels of noise, and for real EEG signals, recorded from an auditory experiment with 2 conditions (Frequent and No Frequent) and using 128 electrodes cap. For both cases, a systematic reduction of the number of electrodes was made based on the electrodes signal power. Using synthetic EEG signals it was found that MFOCUSS and MSP produces localizations with lower error medias. Even more, MFOCUSS localized sources with no error in many cases and MSP obtained sources positions with approximately 2 cm of error. These results were consistent even for signals with 10 dB of SNR and 32 electrodes. In the analysis of contrasts between more active sources (t-test with p<0.05 and FEW correction) found for each condition (No Frequent > Frequent) from the real EEG signals it was noted that MSP presented consistent results when reducing the number of electrodes to 32.
Este ítem está sujeto a una licencia Creative Commons Licencia Creative Commons