Browsing by Author "Hecker, Kenton"

Now showing 1 - 1 of 1
  • Thumbnail Image
    ItemOpen Access
    EMD-Based EEG and fMRI Data Analysis and Integration for High Precision Brain Functional Imaging
    (2023-04-28) Moradi, Narges; Sotero Diaz, Roberto; Gordon Goodyear, Bradley; Protzner, Andrea; Wiebe, Samuel; Leung, Henry; Murari, Kartikeya; Hecker, Kenton; Meltzer, Jed
    To understand the spatiotemporal dynamics of brain function, both high spatial and temporal resolution imaging data are required. Electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) are two non-invasive and complementary functional brain imaging methods with high temporal and spatial resolution, respectively. Thus, combining EEG and fMRI data would permit mapping of brain function with high temporal and spatial resolution simultaneously. The accurate design of models for EEG-fMRI data integration has become an important research topic; however, to date, the success of these models has been limited. In this thesis, new analysis methods are proposed that combine mathematically decomposed components of EEG and fMRI data, guided by information about the underlying neural activity. Specifically, this thesis introduces and validates novel methods based on Empirical Mode Decomposition (EMD) to remove the global signal (GS) from fMRI, denoise the gamma frequency band of EEG, and to integrate EEG with fMRI data. EMD-based methods, ICEEDMAN and the 3D-EMD, are used to decompose EEG and fMRI data into temporal- and spatial-Intrinsic Mode Functions (IMFs) (TIMFs and SIMFs, respectively). First, we show that GS can be removed from the fMRI by removing low-frequency SIMFs causing spurious high global connectivity in the brain. Second, we denoise EEG Gamma-band by removing low-power TIMFs from its frequency- and amplitude-modulated components corresponding to the noise. Finally, we improve EEG source localization precision by adding fMRI’s high spatial-frequency-based weights to the EEG inverse problem’s gain matrix, thereby improving EEG's spatial resolution, especially when deep regions are involved. This thesis thus develops novel methods to increase the precision of imaging of brain function with less artifacts and high spatial and temporal resolutions. A more complete representation of brain function is crucial for better brain function realization, accurate diagnosis, and development of effective treatments for brain diseases such as epilepsy and ADHD. It could aid by specifying sources of seizures and distinguishing patterns and networks involved in ADHD and epileptic activities with high precision.