Institutional Repository
Technical University of Crete
Technical University of Crete
EN
|
EL
| URI: | http://purl.tuc.gr/dl/dias/FFBC94AE-F765-4176-9777-1F12B6325700 | ||
| Year | 2021 | ||
| Type of Item | Diploma Work | ||
| License |
|
||
| Bibliographic Citation | Glykeria Sdoukopoulou, "Band specific oscillations in combined EEG/MEG source analysis: Case study in drug resistant epilepsy ", Diploma Work, School of Electrical and Computer Engineering, Technical University of Crete, Chania, Greece, 2021 https://doi.org/10.26233/heallink.tuc.88396 | ||
| Appears in Collections |
Objective: In the current study, we investigate the contribution of HighFrequency Oscillations (HFOs) on the indication of the epileptogenic zone(EZ). We also study new approaches for the automatic detection of the noncerebral activity.Motivation: We deployed an integrated pipeline for the detection of theEZ, incorporating HFOs-based and interictal spikes-based source analysis ona multi-focal epilepsy case. HFOs have been shown similar and/or betteraccuracy on the indication of the EZ compared to interictal spikes. Moreover,a combination of the non-invasive modalities electro- and magnetoencephalography (EEG) and (MEG), EMEG, has been shown to outperform single EEG or MEG in source analysis.Novelty: In this thesis, we provide a patient-specific pipeline to investigateHFOs contribution on source localization compared to annotated interictalspikes. We have detected scalp HFOs on each modality using a thresholdingtechnique in combination with an energy-based clustering approach. A calibrated realistic FEM head modelling is used to implement HFOs-based andinterictal spikes-based source localization, independently.Methods: The brain activity is recorded by EEG and MEG on a patientwho suffered from multi-focal epilepsy. The first step for the EEG/MEGprocessing was the preprocessing, including filtering and artifact detectionand correction methods in combination with information theory metrics. Wealso investigate new approaches towards the restriction of the non cerebralactivity. Consequently, scalp HFOs are detected on both EEG and MEG.The detection algorithm consists of 3-phases sequentially implemented incorporating a thresholding technique, visual inspection and energy-based clustering approach. Source analysis is performed on detected HFOs and annotated spikes. For each epileptic indicator, a solution to source localization iscalculated, using the sLORETA algorithm, for different time instances andfor each modality. A realistic head model including six tissue compartments,white matter anisotropy and calibrated skull conductivities has been used. Acomparison between HFOs and interictal spikes-based source reconstructionsis performed.Results and Conclusions: Independently for HFOs and interictal spikes,four time instances have been chosen for each modality to explain better theunderlying epileptic activity and the propagation phenomenon of the activitybetween the lesions. The results show that there is a concordance between the HFOs-based and spikes-based source estimation in all modalities. Although,single EEG and MEG source analysis indicates successfully the first lesschallenging lesion, their source reconstructions are far away from the secondlesion. EMEG source reconstruction is able to indicate both lesions, revealingalso a pathway between them, especially when using HFOs. Therefore, bi-lesional epileptic activity can be detected through the synergy of EEG andMEG with HFOs on the basis of realistic head model.
Copyright © DIAS 2013
