fMRI Connectivity: Mechanisms & Analysis

Monday 22 April 2013

Iris Y. Zhou^1,2, Y. X. Liang³, Russell W. Chan^1,2, Shujuan Fan^1,2, Patrick P. Gao^1,2, Joe S. Cheng^1,2, K. F. So³, and Ed X. Wu^1,2
1Laboratory of Biomedical Imaging and Signal Processing, The University of Hong Kong, Hong Kong SAR, China, ²Department of Electrical and Electronic Engineering, The University of Hong Kong, Hong Kong SAR, China, ³Department of Anatomy, The University of Hong Kong, Hong Kong SAR, China

This study explored the role of anatomical/axonal connections in resting-state fMRI connectivity and the plasticity of resting-state networks. Animal models of complete and partial corpus callosum (CC) transection were studied with rsfMRI in conjunction with intracortical EEG recording and Mn2+ tracing of axonal connections. At post-surgery day 7, resting-state connectivity significantly decreased in the cortical areas whose callosal connections were severed. At post-surgery day 28, disrupted connectivity was partly restored in partial transection group, likely through the spared pathways in remaining CC. These rsfMRI findings were paralleled by EEG recording and. Mn2+ tracing results. These results directly support the primary and indispensable role of anatomical/axonal connections via CC in resting-state fMRI connectivity, and that anatomical connection based resting-state networks can be plastic.

Chi Wah Wong¹, Valur Olafsson¹, Omer Tal¹, and Thomas Liu^1
1Center for Functional MRI, University of California San Diego, La Jolla, CA, United States

A prior study has shown that caffeine reduces the amplitude of the global signal in resting-state fMRI and enhances the anti-correlation between the Default Mode Network (DMN) and Task Positive Network (TPN). In this study, we used simultaneous EEG-fMRI to investigate the neural-electrical basis of these caffeine-related effects. We found that the caffeine-induced changes in the global signal amplitude are negatively correlated with changes in vigilance derived using EEG.

Thomas Allan¹, Matthew J. Brookes¹, Susan T. Francis¹, and Penelope A. Gowland^1
1SPMMRC, University of Nottingham, Nottingham, Nottinghamshire, United Kingdom

The origins of functional connectivity are unknown but contributions from spontaneous neural events may have an effect of typical measures of functional connectivity. These spontaneous events, whether they are externally stimulated or internally driven can evoke bilateral responses in multiple network nodes. Here we show that spontaneous BOLD events, detected using paradigm free mapping, have a significant contribution to functional connectivity on a short window correlation analysis and that these events are clustered in space and do not require the entire network to perform a task, highlighting substructures within large scale networks.

Robert Smith^1
1Neurology, UCLA, Los Angeles, CA, United States

The present study explores multi-scale entropy (MSE) analysis to investigate the entropy of resting state fMRI signals across multiple time scales. MSE analysis distinguishes random noise from complex signals since the entropy of the former decreases with larger time scales while the latter signal maintains its entropy due to “self-similarity" across time scales. The results show enhanced contrast in entropy between gray and white matter, as well as between age groups using MSE analysis.

Hao Jia¹, Xiaoping P. Hu², and Gopikrishna Deshpande^1,3
1AU MRI research center, ECE dept., Auburn University, Auburn, AL, United States, ²Coulter Dept. of Biomedical Engineering, Georgia Institute of Technology & Emory Univeristy, Atlanta, GA, United States, ³Dept. of Psychology, Auburn University, Auburn, AL, United States

We assume connectivity dynamics derived from fMRI have finite, quasi-stable configurations based on previous EEG/fMRI evidence. We tested this using a unified framework involving dynamic estimation of whole brain functional connectivity (FC) and effective connectivity (EC), evolutionary clustering and segmentation into finite number of patterns. Sliding window method was used to determine FC and dynamic granger method was used for EC. Result evidenced above hypothesis and there are 2-3 dominant modes for both FC and EC. Main FC modes feature default mode network, visual, sub-cortical and motor networks, while sensory regions to frontal cortex interaction is revealed by EC modes.

Sreenath Pruthviraj Kyathanahally¹, Nurhan Erbil¹, Vince D. Calhoun^2,3, and Gopikrishna Deshpande^1,4
1AU MRI Research Center, Department of Electrical and Computer Engineering, Auburn University, Auburn, Alabama, United States, ²The Mind Research Network and Lovelace Biomedical and Environmental Research Institute, Albuquerque, NM, United States, ³Department of Electrical and Computer Engineering, University of New Mexico, Albuquerque, Alabama, United States, ⁴Department of Psychology, Auburn University, Auburn, Alabama, United States

Previous work using simultaneous EEG/fMRI has shown that the slow temporal dynamics of resting state networks (RSNs) obtained from fMRI are correlated with smoothed and down sampled versions of various EEG features such as microstates and band-limited power and these RSNs not only exist in the low frequency but also exist in high frequency. In this study, to test this critical hypothesis, we acquired a simultaneous EEG/fMRI from which involved fast fMRI sampling using multiband EPI and EPI acquisition. We found DMN in both MB-EPI and EPI acquisition however the correlation coefficient between two modalities in MB-EPI was higher.

Karen J. Mullinger¹, Stephen D. Mayhew², Andrew P. Bagshaw², Richard W. Bowtell¹, and Susan T. Francis^1
1SPMMRC, School of Physics and Astronomy, University of Nottingham, Nottingham, United Kingdom, ²BUIC, School of Psychology, University of Birmingham, Birmingham, United Kingdom

Intrinsic connectivity networks (ICNs) exist during both stimulation and rest, however little is known about the changes in their activity in the transition between task and rest. Simultaneous EEG-BOLD measures provide an interesting new method to study this transition. We demonstrate that during median-nerve stimulation BOLD modulations, indexed by EEG mu power, occur over a number of ICNs, whilst post-stimulus modulations are specific to the bilateral sensorimotor network. We hypothesize that the post-stimulus activity represents re-setting of the entire sensorimotor network, providing a mechanism for this ICN to return to resting-state activity from the lateralised activity driven by stimulation.

Xiao Liu¹, Catie Chang¹, and Jeff H. Duyn^1
1Advanced MRI section, LFMI, NINDS, National Institutes of Health, Bethesda, MD, United States

In this study, we identified 30 spontaneous co-activation patterns (CAPs) by regrouping and then averaging resting-state fMRI volumes. The CAPs present interesting information regarding spontaneous brain activity at distinct time points: e.g., multiple default mode network (DMN) CAPs with distinct features, multiple “task-positive” CAPs anti-correlated with DMN region, very specific thalamocortical connections, and altered occurrence rates in different populations. The data-driven approach used here may serve as a novel method for analyzing and interpreting resting-state fMRI signals, complementary to conventional approaches.

Senhua Zhu^1,2, Zhuo Fang^1,2, Siyuan Hu³, Marc Korczykowski², Ze Wang², John A. Detre², and Hengyi Rao^1,2
1Psychology, Sun Yat-sen University, Guangzhou, Guangdong, China, ²Center for Functional Neuroimaging, University of Pennsylvania, Philadelphia, PA, United States, ³State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University, Beijing, Beijing, China

The present study used independent component analysis to examine resting brain networks in a large cohort (n=149) of subjects with arterial spin labeling (ASL) perfusion MRI data. Ten CBF networks were consistently identified across the whole and sub-datasets, including the default mode network, bilateral attention networks, primary and second visual networks, auditory network, ventral-medial prefrontal network, dorsal-medial prefrontal network, and two limbic networks. These networks well replicated the resting-state BOLD networks from a sub-group (n=81) and support the feasibility of using CBF connectivity to examine resting brain function.

Ying-Hua Chu¹, Jyrki Ahveninen², Tommi Raij², Wen-Jui Kuo³, John W. Belliveau², and Fa-Hsuan Lin^1
1Institute of Biomedical Engineering, National Taiwan University, Taipei, Taiwan, ²A. A. Martinos Center, Massachusetts General Hospital, Charlestown, MA, United States,³Institute of Neuroscience, National Yang-Ming University, Taipei, Taiwan

Resting-state fMRI studies reflecting functional connectivity have been typically limited to frequencies below 0.1 H. Here, we hypothesize that fMRI can detect interregional correlations in MRI time series at frequencies above 0.1 Hz as well. Using MR inverse imaging (InI) at a 10 Hz sampling rate, we studied interhemispheric correlations between primary sensorimotor and visual cortices. We found significant correlations at 4 Hz (average Z-score ~8) that were about 60% of those observed at 0.1 Hz.