What Is an EEG fMRI?

The study of the internal working mechanism of the brain is a major challenge in the 21st century. Functional neuroimaging has an excellent application prospect in understanding brain functions and disorders (Friston, 2009), and has now become an important tool for neuroscience research and clinical applications. Noninvasive high-time, high-spatial-resolution imaging is the tool that brain research is looking forward to. It is helpful to understand the mechanisms of human attention, execution and memory, and to reveal mechanisms such as the origin and transmission of epilepsy. At present, the main non-invasive technologies of the human brain include magnetic resonance imaging (MRI), electroencephalogram (EGG), and magnetoencephalogram (MEG). They have their own advantages and disadvantages in terms of space-time resolution. Combining different neuroimaging modes (multimodal imaging, especially EEG-MRI fusion) may not only achieve complementary advantages and improve spatiotemporal resolution (Figure 1), but also help our understanding of neural activity in a single modality (Biessmann, Plis, Meinecke, Eichele, & Muller, 2011; Huster, Debener, Eichele, & Herrmann, 2012; Laufs, 2012) to advance our understanding of the pathophysiology of neurological and mental illness.

Chinese name
Magnetic resonance imaging
Foreign name
Magnetic Resonance Image, MRI

EEG-MRI fusion background

The study of the internal working mechanism of the brain is a major challenge in the 21st century. Functional neuroimaging has an excellent application prospect in understanding brain functions and disorders (Friston, 2009), and has now become an important tool for neuroscience research and clinical applications. Noninvasive high-time, high-spatial-resolution imaging is the tool that brain research is looking forward to. It is helpful to understand the mechanisms of human attention, execution and memory, and to reveal mechanisms such as the origin and transmission of epilepsy. At present, the main non-invasive technologies of the human brain include magnetic resonance imaging (MRI), electroencephalogram (EGG), and magnetoencephalogram (MEG). They have their own advantages and disadvantages in terms of space-time resolution. Combining different neuroimaging modes (multimodal imaging, especially EEG-MRI fusion) may not only achieve complementary advantages and improve spatiotemporal resolution (Figure 1), but also help our understanding of neural activity in a single modality (Biessmann, Plis, Meinecke, Eichele, & Muller, 2011; Huster, Debener, Eichele, & Herrmann, 2012; Laufs, 2012) to advance our understanding of the pathophysiology of neurological and mental illness.
Figure 1: Motivation for integration: complementary advantages through each modality.

- EEG-magnetic resonance imaging fusion

Multi-modal imaging is mainly to integrate different measurement modes of neural activity. Among them, EEG, as a non-invasive method of recording electrophysiological signals of brain activity, has a high temporal resolution (millisecond level), but a low spatial resolution; while another non-invasive technique currently commonly used, MRI is just the opposite. Obviously, the combination of these two non-invasive recording methods has the potential to produce comprehensive neuroimaging technology with high spatiotemporal resolution. Its superior spatio-temporal resolution will be beneficial for studying cognitive-related mechanisms of brain function and neurological and psychiatric diseases. This article only introduces the fusion of structural information [structural MRI (sMRI), diffusion tensor imaging (DTI)] and functional information (EEG, fMRI). It is worth noting that the collection of structural information is relatively standard; and the synchronous acquisition of EEG-fMRI requires simultaneous EEG recording and fMRI scanning under a compatible acquisition system, so that the neuroelectrophysiological activity and All metabolic activities were recorded (Figure 2).
Figure 2: EEG-fMRI synchronous acquisition system (Laufs, Daunizeau, Carmichael, & Kleinschmidt, 2008)

1- EEG-magnetic resonance imaging fusion 1 EEG-magnetic resonance imaging information fusion

EEG-sMRI / DTI fusion
The fusion of EEG and MRI is firstly reflected in the fact that sMRI provides the head volume conductor model for EEG source localization, which is extremely important for the accuracy of source imaging. Based on sMRI, a boundary element model, a finite element model, or a finite voxel model of individual features can be constructed. In the fusion of EEG and DTI data, the existing research mainly uses the information of neuron rhythm activity obtained from EEG data and the fiber bundle or dispersion obtained from DTI data to analyze and explore the relationship between the rhythm of the brain and the structural basis (Valdes-Hernandez et al ., 2010; Whitford et al., 2011). On the whole, the fusion of EEG with high temporal resolution and sMRI / DTI with high spatial resolution to explore the relationship between brain function and structure is still a very promising and challenging subject.
EEG-fMRI fusion
Regarding the fusion of EEG and fMRI, there are currently many different fusion methods (Biessmann, et al., 2011; Huster, et al., 2012; Rosa, Daunizeau, & Friston, 2010). EEG-fMRI fusion is usually divided into the following three categories: 1. spatial information fusion, 2. temporal information fusion, 3. model / symmetric fusion. As shown in Figure 3.
Figure 3: Three ideas of fusion: (i). Fusion of spatial information; (ii). Fusion of temporal information; (iii). Symmetrical fusion of space-time information.
(i) Spatial information fusion
A common method for spatial integration is to use fMRI activation maps as a priori information on the location of the EEG source. Existing methods include fMRI-constrained dipole localization methods and fMRI-constrained / weighted distributed source imaging methods. Due to the increase of fMRI space constraints, the ill-posedness of the EEG inverse problem can be alleviated to some extent. In recent years, some research groups have also innovatively introduced the fMRI function network obtained by independent component analysis as the prior information of EEG source positioning into the EB (empirical Bayes) model of EEG source estimation, and developed a signal based on fMRI network. Constrained EEG source localization technology (Lei et al., 2010).
(ii) Time information fusion
Integration in the time domain provides information for functional magnetic resonance imaging by using EEG dynamic signal characteristics in the time or frequency domain. These characteristic information usually come from data of a specific time period or a specific frequency band. Usually, the variables obtained from this electromagnetic record are convolved with the conventional hemodynamic response function (HRF), and then linear regression (general liner model, GLM) is performed with the voxel-based BOLD signal to determine the corresponding FMRI statistical activation map of the electromagnetic signal characteristics of interest. Lei et al. Developed a method for inversion of HRF functions based on empirical Bayes (EB) models (Lei, Qiu, Xu, & Yao, 2010). This technology uses the high temporal resolution neural activity time information and intensity information provided by EEG to reconstruct the hemodynamic function of the corresponding fMRI.
(iii) Model symmetry fusion
The two fusion analysis methods above focus on the help of one mode to the other. In addition, EEG-fMRI fusion also includes fusion methods driven by "genesis model driving" or "signal model driving". A typical example is DCM (Friston, Harrison, & Penny, 2003; Kiebel, Garrido, & Friston, 2007). Because model-driven fusion is usually based on neurophysiology, the inversion of these models will help us understand the origin of neural activity and the corresponding regulatory mechanisms, but the construction of models is more difficult. To this end, some researchers have also studied the interdependence between EEG and fMRI signals, that is, the use of mutual information to achieve the fusion of EEG and fMRI (Valdes-Sosa et al., 2009). Solve the inverse space and time problems of the EEG and BOLD signals respectively, and then use the maximum correlation of the time characteristics of the two to achieve fusion.
In summary, in terms of simultaneous EEG-fMRI fusion, there have been many methods and techniques to integrate EEG and fMRI. However, the credibility of the integrated information is not yet clear. Recently, based on existing research, Dong et al. Proposed a hierarchical framework to distinguish the credibility of the integration results, and used this hierarchical framework to understand brain activity (Dong et al., 2014). Only by matching the two aspects of time and space can we get the most reliable information with high spatiotemporal resolution. Simulation experiments and real data verification show that the framework is a feasible solution for explaining the cognitive process of the human brain.

IN OTHER LANGUAGES

Was this article helpful? Thanks for the feedback Thanks for the feedback

How can we help? How can we help?