Introduction to resting state fMRI and functional connectivity analysis

Motivations for resting-state fMRI and the study of functional connectivity

Functional MRI (fMRI) is a technique used to study ongoing large-scale neural dynamics. fMRI is measuring the blood oxygen level dependent (BOLD) signal, which reflects local changes in blood flow that is coupled with underlying neural activity through neurovascular coupling mechanisms (Logothetis, 2008). fMRI has been initially mainly used in task paradigms where participants undergo scanning while performing a task, and the contrast in BOLD signal during task compared to baseline is evaluated in every voxel through a general linear model (GLM) approach. Yet, in the last 20 years there has been an ever-increasing interest in studying BOLD dynamics during resting-state, termed resting-state fMRI (rs-fMRI). This interest stems largely from the observation that even during rest, distinct brain regions can display coordinated activity over time, termed functional connectivity (FC). Following the initial observation of this phenomenon in the bilateral primary motor cortices by Biswal and colleagues (Biswal et al., 1995), further investigations demonstrated that brain areas tend to group into various reproducible large-scale brain "networks" of coactivated regions (Beckmann et al., 2005; Greicius et al., 2003; Yeo et al., 2011). This observation is believed to reflect distributed integrative systems required for the regulation of complex brain functions(Yeo et al., 2011). rs-fMRI FC research is thus aimed at investigating how properties of large-scale brain dynamics inform us about the regulation of brain function, and how these measures of brain dynamics can be developed into phenotypic characterization of individuals’ brain function (Kelly et al., 2012).

Main assumptions for the rs-fMRI FC pursuit and its criticisms

1. the BOLD signal is reflecting underlying neural activity -> it doesn’t always, and it is a constant challenge to distinguish neural from non-neural contributions (a good critical reading on this is (Logothetis 2008))
2. the identified co-activation patterns across brain regions (i.e. functional connectivity) are reflective of ongoing information processing events relevant for brain function
3. reliable phenotypes can be characterized at the individual with an fMRI scan -> challenge of distinguishing individual differences driven by neural dynamics as opposed to other possible confounding sources (motion, arousal, vasculature,...), and also whether the timeframe (e.g. 6-10min scan) of recording is sufficient to map reliable phenotypes

The basis for fMRI data: from EPI acquisition to the voxel timeseries

fMRI scan is conducted based on echo-planar imaging (EPI) MRI sequences. This imaging technique repeatedly acquires T2-weighted slices covering the whole brain within a very short time intervals (repetition time (TR) of 1-2sec). The resulting EPI scan is a 4D image representing a sequence of 3D brain volumes acquired over time at an interval corresponding to the TR.
The decay of protons in the transverse x-y plane (the T2 decay) occurs much faster than the T1 decay given that protons become out of phase in their spin along the transverse plane, which is referred to as the T2* effect. Deoxygenated hemoglobin is paramagnetic and will thus disturb local magnetic properties in tissues and change the speed of T2 decay resulting from the T2* effect (which is reflected in our images as changes in voxel intensity). This renders the EPI image sensitive to local changes in deoxyhemoglobin concentration, which forms the basis of the BOLD signal. The BOLD signal is related to underlying neural activity through mechanisms of neurovascular coupling, where the regulation of blood flow across the brain controls the distribution of oxygenated blood to follow changes in metabolic activity in the neural tissue (for a review on neurovascular coupling see (Iadecola, 2017)).
After proper preprocessing of the EPI image, we can derive the intensity values over time for a given voxel, termed voxel timeseries, to observe changes in the BOLD signal in the given anatomical region over time. This forms the basis for any subsequent analysis using fMRI.

Main analysis approaches (or first-level analysis):

• Seed-based FC: this technique is the most simple analysis of FC, and corresponds in the correlation between the mean timeseries within a given anatomical (i.e.a set of labelled voxels in an anatomical region of interest (ROI)) and every other voxel timeseries. This results in a brain correlation map for a given seed.

  • Advantages: 1-interpretability: it gives a anatomical map which can be visualized, and artefactual correlations are usually straightforward to distinguish (see example below) 2-data quality control: can verify whether an expected brain network is appropriately mapped to validate the quality of the data

  • Disadvantage: 1-not data-driven: the analysis is concentration only a the single anatomical region which was given a seed 2-’static’ measure of FC: the relationship of two regions is simplified into a single measure of connectivity (see dynamic FC for alternatives)

• Whole-brain FC matrix, or ‘connectome’: this technique is an extension of the seed-based FC approach, where instead the whole brain anatomy is parcellated into a set of ROIs, and the pairwise correlation coefficients of every ROIs’ timeseries is computed. This can also be conducted at the voxel timeseries level to obtain a voxelwise instead of a parcellated connectome. The correlations can be represented in a ROI-by-ROI matrix format to give a wholistic view of the brain connectivity (see image attached).
  • Advantages: 1-data-driven 2-wholistic representation of connectivity
  • Downside: 1-the sources of the correlations are hidden (e.g. can’t easily distinguish when confounds are driving the correlations, or can’t confirm whether there is sufficient BOLD signal to derive meaningful measures) 2-bias of the parcellation used 3-’static’ measure of FC: same as seed-based FC

• spatial-Independent component analysis (ICA): this approach consists of using ICA, a dimensionality reduction technique, to decompose the variation in BOLD signal across the brain into distinct sources. This technique effectively identifies distinct spatial components that may correspond to neural as well as noise sources. The neural-related components will identified collections of regions that tend to be co-activated together, or ‘brain networks’. Noise-related components (usually >80% of the components) will display distinct characteristics related to motion or non-neural physiological sources (can read (Griffanti et al., 2017) (humans) or (Zerbi et al., 2015) (mice) for more details on the identification of components)
  • Advantages: 1-interpretability is straightforward (looking at spatial maps and their associated timeseries) 2-distinguishes neural sources from noise sources 3-completely data-driven
  • Downside: 1-somewhat ‘static’ measure of connectivity since it only identifies spatial relationships without regards for the temporal structure of the BOLD signal

Alternative analysis approaches:

• co-activation patterns (CAPs) (Liu & Duyn, 2013; Liu et al., 2018) (a recently published toolbox https://www.sciencedirect.com/science/article/pii/S1053811920301087)
• Quasi-periodic patterns (QPPs) (Belloy et al., 2018; Thompson et al., 2014)
• Functional gradients (Margulies et al., 2016)
• sliding-window FC (Allen et al., 2014)
• Wavelets time-frequency analysis (Chang & Glover, 2010)

Suggested readings

• Neurovascular coupling (Iadecola, 2017)
• On the interpretation of the BOLD signal (Logothetis, 2008)
• Conservation of brain networks during rest and task (Smith et al., 2009)
• Concerns with false positives in task fMRI (Cluster failure) (Eklund et al., 2016)
• Review on dynamic FC (Hutchison et al., 2013)

References

• Allen, E. A., Damaraju, E., Plis, S. M., Erhardt, E. B., Eichele, T., & Calhoun, V. D. (2014). Tracking whole-brain connectivity dynamics in the resting state. Cerebral Cortex , 24(3), 663–676.
• Beckmann, C. F., DeLuca, M., Devlin, J. T., & Smith, S. M. (2005). Investigations into resting-state connectivity using independent component analysis. Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences, 360(1457), 1001–1013.
• Belloy, M. E., Naeyaert, M., Abbas, A., Shah, D., Vanreusel, V., van Audekerke, J., Keilholz, S. D., Keliris, G. A., Van der Linden, A., & Verhoye, M. (2018). Dynamic resting state fMRI analysis in mice reveals a set of Quasi-Periodic Patterns and illustrates their relationship with the global signal. NeuroImage, 180(Pt B), 463–484.
• Biswal, B., Zerrin Yetkin, F., Haughton, V. M., & Hyde, J. S. (1995). Functional connectivity in the motor cortex of resting human brain using echo-planar MRI. Magnetic Resonance in Medicine: Official Journal of the Society of Magnetic Resonance in Medicine / Society of Magnetic Resonance in Medicine, 34(4), 537–541.
• Chang, C., & Glover, G. H. (2010). Time-frequency dynamics of resting-state brain connectivity measured with fMRI. NeuroImage, 50(1), 81–98.
• Eklund, A., Nichols, T. E., & Knutsson, H. (2016). Cluster failure: Why fMRI inferences for spatial extent have inflated false-positive rates. Proceedings of the National Academy of Sciences of the United States of America, 113(28), 7900–7905.
• Greicius, M. D., Krasnow, B., Reiss, A. L., & Menon, V. (2003). Functional connectivity in the resting brain: a network analysis of the default mode hypothesis. Proceedings of the National Academy of Sciences of the United States of America, 100(1), 253–258.
• Griffanti, L., Douaud, G., Bijsterbosch, J., Evangelisti, S., Alfaro-Almagro, F., Glasser, M. F., Duff, E. P., Fitzgibbon, S., Westphal, R., Carone, D., Beckmann, C. F., & Smith, S. M. (2017). Hand classification of fMRI ICA noise components. NeuroImage, 154, 188–205.
• Honey, C. J., Sporns, O., Cammoun, L., Gigandet, X., Thiran, J. P., Meuli, R., & Hagmann, P. (2009). Predicting Resting-State Functional Connectivity from Structural Connectivity. Proc. Intl. Soc. Mag. Reson. Med, 17, 859.
• Hutchison, R. M., Womelsdorf, T., Allen, E. A., Bandettini, P. A., Calhoun, V. D., Corbetta, M., Della Penna, S., Duyn, J. H., Glover, G. H., Gonzalez-Castillo, J., Handwerker, D. A., Keilholz, S., Kiviniemi, V., Leopold, D. A., de Pasquale, F., Sporns, O., Walter, M., & Chang, C. (2013). Dynamic functional connectivity: promise, issues, and interpretations. NeuroImage, 80, 360–378.
• Iadecola, C. (2017). The Neurovascular Unit Coming of Age: A Journey through Neurovascular Coupling in Health and Disease. Neuron, 96(1), 17–42.
• Kelly, C., Biswal, B. B., Craddock, R. C., Castellanos, F. X., & Milham, M. P. (2012). Characterizing variation in the functional connectome: promise and pitfalls. Trends in Cognitive Sciences, 16(3), 181–188.
• Liu, X., & Duyn, J. H. (2013). Time-varying functional network information extracted from brief instances of spontaneous brain activity. Proceedings of the National Academy of Sciences of the United States of America, 110(11), 4392–4397.
• Liu, X., Zhang, N., Chang, C., & Duyn, J. H. (2018). Co-activation patterns in resting-state fMRI signals. NeuroImage, 180(Pt B), 485–494.
• Logothetis, N. K. (2008). What we can do and what we cannot do with fMRI. Nature, 453(7197), 869.
• Margulies, D. S., Ghosh, S. S., Goulas, A., Falkiewicz, M., Huntenburg, J. M., Langs, G., Bezgin, G., Eickhoff, S. B., Castellanos, F. X., Petrides, M., Jefferies, E., & Smallwood, J. (2016). Situating the default-mode network along a principal gradient of macroscale cortical organization. Proceedings of the National Academy of Sciences of the United States of America, 113(44), 12574–12579.
• Smith, S. M., Fox, P. T., Miller, K. L., Glahn, D. C., Fox, P. M., Mackay, C. E., Filippini, N., Watkins, K. E., Toro, R., Laird, A. R., & Beckmann, C. F. (2009). Correspondence of the brain’s functional architecture during activation and rest. Proceedings of the National Academy of Sciences of the United States of America, 106(31), 13040–13045.
• Thompson, G. J., Pan, W.-J., Magnuson, M. E., Jaeger, D., & Keilholz, S. D. (2014). Quasi-periodic patterns (QPP): large-scale dynamics in resting state fMRI that correlate with local infraslow electrical activity. NeuroImage, 84, 1018–1031.
• Yeo, B. T. T., Krienen, F. M., Sepulcre, J., Sabuncu, M. R., Lashkari, D., Hollinshead, M., Roffman, J. L., Smoller, J. W., Zöllei, L., Polimeni, J. R., Fischl, B., Liu, H., & Buckner, R. L. (2011). The organization of the human cerebral cortex estimated by intrinsic functional connectivity. Journal of Neurophysiology, 106(3), 1125–1165.
• Zerbi, V., Grandjean, J., Rudin, M., & Wenderoth, N. (2015). Mapping the mouse brain with rs-fMRI: An optimized pipeline for functional network identification. NeuroImage, 123, 11–21.