Adaptive Cortical Parcellations for Source Reconstructed EEG/MEG Connectomes
Presented During:
Tuesday, June 27, 2017: 11:32 AM - 11:45 AM
Vancouver Convention Centre
Room:
Ballroom AB
Submission No:
1790
Submission Type:
Abstract Submission
On Display:
Monday, June 26 & Tuesday, June 27
Authors:
Seyedehrezvan Farahibozorg1,2, Richard Henson2, Olaf Hauk2
Institutions:
1University of Cambridge, Cambridge, United Kingdom, 2MRC Cognition and Brain Sciences Unit, Cambridge, United Kingdom
First Author:
Seyedehrezvan Farahibozorg
-
Lecture Information
|
Contact Me
University of Cambridge|MRC Cognition and Brain Sciences Unit
Cambridge, United Kingdom|Cambridge, United Kingdom
University of Cambridge|MRC Cognition and Brain Sciences Unit
Cambridge, United Kingdom|Cambridge, United Kingdom
Introduction:
There is growing interest in the rich temporal and spectral properties of Electro- and Magnetoencephalography (E/MEG) signals in order to study the functional connectome of the brain [1, 2]. However, the spatial resolution of E/MEG data is limited, because several thousand sources of activation in the brain must be estimated from maximally a few hundred recording sites. This limited spatial resolution causes the so-called leakage problem: activity estimated in one region of interest (ROI) can be affected by leakage from locations outside this ROI [3, 4]. E/MEG studies typically adopt parcellations from structural or fMRI research for whole-brain connectivity analysis [5]. However, considering the spatial resolution of E/MEG, these parcellations are unlikely to be optimal [6]. Here, we utilise Cross-Talk Functions (CTFs) as a direct measure of spatial leakage [7] and utilise two CTF-informed image segmentation algorithms in order to parcellate the cortical surface into the maximum number of distinguishable ROIs.
Methods:
We computed resolution matrices (with rows as CTFs) for individual subjects, based on forward and inverse models computed using BEM head models and L2 MNE inverse operators of 17 healthy subjects. In the first parcellation approach, we started from standard anatomical parcellations and modified the ROIs using a CTF-informed split-and-merge (SaM) algorithm [8]. In the second approach, we started from all brain vertices with no prior parcellation. A CTF-informed region growing (RG) algorithm [8] was used to create ROIs around the vertices that showed highest sensitivity and specificity of CTFs on the cortex, which were then optimised using an SaM algorithm. The algorithms are designed such that they merge ROIs/vertices with highly overlapping CTFs, split ROIs that produce distinguishable patterns of CTFs, remove ROIs with low sensitivity, and for each ROI identify a group of representative vertices that show high sensitivity and specificity to that particular ROI. We used ROI Resolution Matrices (RRmat) to quantify leakage from each ROI to all other ROIs in the brain in order to evaluate the parcellations' performance where an ideal RRmat is an identity matrix. Thereafter, we evaluated the possible consequences of using different parcellation methods for graph-theoretical connectivity analyses on simulated data with realistic levels of noise.
Results:
Based on the RRmats (Fig. 1), we found that parcellation sensitivity improved from 0.47 and 0.37 in two standard anatomical parcellations (Desikan-Killiany (DKA) and Destrieux Atlases (DA) respectively) to 0.65, 0.70 and 0.70 in modified DKA, DA and RG parcellations respectively. Moreover, ROI distinguishability improved from 0.50 and 0.38 to 0.61, 0.65 and 0.64 (Fig. 1). Interestingly, in spite of their different starting points, both SaM and RG algorithms yielded approximately 70 ROIs. Furthermore, our simulated realistic connectome with a single hub showed that modified parcellations were particularly successful in improving hub sensitivity and hub connectivity probability patterns (Fig. 2).
Conclusions:
Our proposed parcellation algorithms significantly improved the sensitivity and distinguishability of ROIs compared to the anatomical parcellations, while at the same time maximising the number of distinguishable ROIs in the brain. The algorithms are adaptive with respect to the measurement configuration and source localisation methods. Regardless of the starting point they yielded around 70 ROIs, suggesting that this reflects the resolution limit of this particular sensor configuration and source estimation method. Furthermore, our simulations showed that the choice of parcellation can have significant impact on the outcome of graph theoretical analysis of the source-reconstructed E/MEG. Therefore, we conclude that adaptive parcellations are essential for whole-brain EEG/MEG connectomics.
Imaging Methods:
EEG
MEG
Informatics:
Brain Atlases
Modeling and Analysis Methods:
EEG/MEG Modeling and Analysis 1
Segmentation and Parcellation 2
Keywords:
Electroencephaolography (EEG)
MEG
Other - Adaptive Parcellation, Cross-Talk Function, Source Localisation, Whole Brain Connectivity, Functional Connectome
1|2Indicates the priority used for review
Would you accept an oral presentation if your abstract is selected for an oral session?
I would be willing to discuss my abstract with members of the press should my abstract be marked newsworthy:
Please indicate below if your study was a "resting state" or "task-activation” study.
By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute the presentation in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels or other electronic media and on the OHBM website.
Healthy subjects only or patients (note that patient studies may also involve healthy subjects):
Internal Review Board (IRB) or Animal Use and Care Committee (AUCC) Approval. Please indicate approval below. Please note: Failure to have IRB or AUCC approval, if applicable will lead to automatic rejection of abstract.
Please indicate which methods were used in your research:
For human MRI, what field strength scanner do you use?
Which processing packages did you use for your study?
Provide references in author date format
[2] Brookes, M.J. et al., 2016. A multi-layer network approach to MEG connectivity analysis. NeuroImage, 132, pp.425–438. Available at: http://dx.doi.org/10.1016/j.neuroimage.2016.02.045.
[3] Schoffelen, J.M. & Gross, J., 2009. Source connectivity analysis with MEG and EEG. Human Brain Mapping, 30(6), pp.1857–1865.
[4] Hauk, O., Wakeman, D.G. & Henson, R., 2011. Comparison of noise-normalized minimum norm estimates for MEG analysis using multiple resolution metrics. NeuroImage, 54(3), pp.1966–1974. Available at: http://dx.doi.org/10.1016/j.neuroimage.2010.09.053.
[5] Colclough, G.L. et al., 2015. A symmetric multivariate leakage correction for MEG connectomes. NeuroImage, 117, pp.439–448. Available at: http://dx.doi.org/10.1016/j.neuroimage.2015.03.071.
[6] Palva, J.M. et al., 2010. Neuronal synchrony reveals working memory networks and predicts individual memory capacity. Proceedings of the National Academy of Sciences of the United States of America, 107(16), pp.7580–5. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2867688&tool=pmcentrez&rendertype=abstract.
[7] Liu, a K., Belliveau, J.W. & Dale, a M., 1998. Spatiotemporal imaging of human brain activity using functional MRI constrained magnetoencephalography data: Monte Carlo simulations. Proceedings of the National Academy of Sciences of the United States of America, 95(15), pp.8945–8950.
[8] Gonzalez , R.C. & Wood, R.E., 2007. Digital Image Processing (3rd Edition),