Source analysis of EEG data from paralysed subjects
- Authors
- Carabali Carabali, Carmen Alicia
- Format
- MasterThesis
- Status
- publishedVersion
- Description
One of the limitations of Electroencephalography (EEG) data is its quality, as it is usually contaminated with electric signal from involuntary muscle movement. This research intends to compare results of different EEG analysis methods applied to EEG recorded in paralysis and in normal conditions during the performance of the oddball task. The aim is to determinate which types of analysis are more appropriate for dealing with EEG data containing myogenic components. The data used for the research is owned by the Brain Signals Laboratory at the Flinders Medical Centre. It is comprised by the recordings of EEG from six subjects during the performance of the oddball task, in normal conditions and during paralysis. The research consisted in perform different forms of analysis to the EEG data and compare the results of the two stages. The software packages used for the analysis were BESA Research 6.0, BESA Statistics and MATLAB code. The analysis performed includes data pre-processing, discrete and distributed source analysis, coherence and frequency analysis. The data recorded during the experiments was pre-processed, then for targets and distractors were extracted. LAURA distributed source analysis was performed and, combined with information from the literature, nine locations were identified as ideal to allocate 9 source dipoles. These 9 dipoles were fitted to the averaged epochs in order to obtain the source waveforms. The waveforms were statistically analysed in order to compare the measured electrical activity before and after the subjects were paralysed. Additionally, connectivity analysis was done using the coherence method available in the software and finally, time-frequency analysis was done applying Welch?s periodogram, short time Fourier transform and continuous wavelet transform. The findings were that distributed source analysis could produce confounded results for EEG contaminated with myogenic signals, conversely, statistical analysis of the waveforms from the discrete source analysis showed that there is not mayor significative difference between paralysis and pre-paralysis source waveforms, therefore, discrete source analysis is recommended as an appropriate method for dealing with EEG data contaminated with myogenic signal. The frequency analysis showed that, effectively, the spectrograms of the source waveforms are very similar in the two stages for the subjects for which the nine dipole montage fitted properly. Regarding to the connectivity results, these were not conclusive and the suggestion was to perform other methods different to coherence in the generated data for obtaining better outcomes.
El presente estudio pretende comparar dos m?todos de an?lisis de se?ales de electroencefalograma, an?lisis de fuente discreto y an?lisis de fuente distribuido, aplicado a datos de encefalograma de seis personas en estado normal y en par?lisis durante la ejecuci?n de la tarea cognitiva Oddball. En el documento se analizan los resultados de los dos m?todos y se eval?a cu?l de los dos tiene mayor efectividad para lidiar con interferencia el?ctrica producida por m?sculo, tambi?n se presenta un an?lisis de coherencia de las se?ales de EEG y finalmente se utiliza an?lisis en el dominio de la frecuencia para seleccionar el mejor m?todo que resulta ser an?lisis de fuente discreta usando un modelo de 9 dipolos el?ctricos.
- Publication Year
- 2014
- Language
- eng
- Topic
- AN?LISIS DE SE?ALES BIOL?GICAS
ELECTROENCEFALOGRAMA
INTERFERENCIA EL?CTRICA
ESPECTRO DE FRECUENCIA
SE?ALES DE ORIGEN MIOG?NICO
- Repository
- Repositorio SENESCYT
- Rights
- openAccess
- License