Resting-state cortex rhythms (RECOR) is a toolbox to estimate the power of EEG rhythms in the eight large-scale brain networks [Lei et al. 2011; Lei 2012]. The EEG forward model is restricted to a high-density canonical cortical mesh, which was extracted from a structural MRI of a neurotypical male in Fieldtrip software (http://fieldtrip.fcdonders.nl/download.php). The mesh has 8,196 vertices, which was uniformly distributed on the gray-white matter interface. Each vertex node is assumed to have one dipole, oriented perpendicular to the surface. The electrodes were registered to the scalp surface. For example, if we have 64 electrodes to record EEG signal, the lead-field would be a matrix with 64 X 8,196. It was calculated within a three-shell spherical head model including scalp, skull, and brain.

RECOR included two steps to calculate the power of EEG rhythms in each brain network. Firstly, network-based source imaging (NESOI) was employed to estimate the cortical sources of EEG rhythms (see Output 2, [Lei et al., 2011]). Eight large-scale brain networks are used as the covariance priors of the EEG source reconstruction using parametric empirical Bayesian. Seven large-scale networks were identified based on 1000 resting-state functional connectivity: visual, somatomotor, dorsal attention, ventral attention, limbic, fronto-parietal, and default networks [Yeo et al., 2011]. Considering the importance of the deep brain structure (thalamus, caudate, hippocampus, amygdala and olfactory), we used the automated anatomical labeling (AAL) parcellation atlas [Tzourio-Mazoyer et al., 2002] to construct the eighth large-scale networks.

The 8,196 vertices were separated to eight subsets based on its nearest neighbor voxel in the large-scale brain network templates. The covariance prior Vi is from the ith brain network and is an 8,196 X 8,196 covariance basis matrix, which is assign the columns and rows with Green function of the mesh adjacency matrix, if their corresponding vertices is involved in ith network, and the other terms with zero [Lei 2011]. The intensity of the neural electric sources of EEG rhythms are iteratively estimated by Restricted maximum likelihood (ReML) algorithm. It has been shown that NESOI was quite efficient when compared to other inverse methods like weighted minimum norm solution, low-resolution brain electromagnetic tomography (LORETA) and multiple sparse prior model (MSP) [Lei et al., 2011].

The second step of RECOR is averaging the solutions of NESOI across all vertices of a given large-scale brain network (see Output 3). Rather than estimation the punctual EEG source patterns of each rhythm, RECOR focused on the large-scale distribution of EEG source and calculated an averaged current density at each network. This is in line with the low spatial resolution of the adopted technique. As the number of electrodes (64) is much lower than that of the unknown current density at each vertex (8,196), solutions of the EEG inverse problem are under-determined and ill conditioned. This averaging step may minimize the effects of poor NESOI estimates in deep brain structure at which the estimation of EEG sources could be imprecise, especially using an EEG spatial sampling about 100 electrodes (10-10 system). In summary, the RECOR software reported the current density both for all the vertices and the eight brain network.

Output 1: Power spectrum and topography

The average power spectrum and topography of EEG rhythms: delta (2–4 Hz), theta (4–8 Hz), alpha1 (8–10.5 Hz), alpha2 (10.5–13 Hz), beta1 (13–20 Hz), beta2 (20–30 Hz), and gamma (30–40 Hz).

Output 2: Current density in the cortex for each rhythm

This is the inverse solution of NESOI.

Output 3: EEG spectral power density for each network

Reference:

1. Xu Lei, Peng Xu, Cheng Luo, Jinping Zhao, Dong Zhou, Dezhong Yao. fMRI Functional Networks for EEG Source Imaging. Hum Brain Mapp. Jul 2011 32(7): 1141-1160. (Cover Article)

2. Xu Lei. Electromagnetic brain imaging based on standardized resting-state networks. Paper presented at: 5th International Conference on Biomedical Engineering and Informatics (BMEI) (Chongqing, China). 2012 40-44.

3. Yeo, B.T., Krienen, F.M., Sepulcre, J., Sabuncu, M.R., Lashkari, D., Hollinshead, M., Roffman, J.L., Smoller, J.W., Zollei, L., Polimeni, J.R., 2011. The organization of the human cerebral cortex estimated by intrinsic functional connectivity. J. Neurophysiol. 106, 1125-1165.

4. Tzourio-Mazoyer, N., Landeau, B., Papathanassiou, D., Crivello, F., Etard, O., Delcroix, N., Mazoyer, B., Joliot, M., 2002. Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain. Neuroimage 15, 273-289.

Publications: (Please email xlei@swu.edu.cn if your work has been published!)

1. Wenrui Zhao, Dong Gao, Faguo Yue, Yanting Wang, Dangdang Mao, Tianqiang Liu, Xu Lei, 2018. Spontaneous Theta Rhythm Predicts Insomnia Duration: A Resting-State EEG Study. In: Delgado-García, J.M., Pan, X., Sánchez-Campusano, R., Wang, R. (Eds.), Advances in Cognitive Neurodynamics (VI). Springer Singapore, Singapore, pp. 359-364.

2. Xu Lei, Taoyu Wu, Pedro Valdes-Sosa. Incorporating priors for EEG source imaging and connectivity analysis. Frontiers in Neuroscience - Brain Imaging Methods. Aug 2015 9: 284.