Signals from the Brain: A Tale of Dynamics and Networks

State-of-the-art neuroimaging such as magnetic resonance imaging (MRI) provides unprecedented opportunities to non-invasively measure human brain structure (anatomy) and function (physiology). To fully exploit the rich spatiotemporal structure of these data and gain insights into brain function in health and disorder, novel signal processing and modeling approaches are needed, instilled by domain knowledge from neuroscience and instrumentation.
I will highlight our main research axes to pursue this endeavor. First, we propose a new framework that leverages sparsity-pursuing hemodynamic deconvolution of functional MRI (fMRI) time series to represent them in terms of transients. Reoccurring spatial configurations of transients then identify a neurologically pertinent repertoire of large-scale distributed patterns. Their temporal dynamics reveal the complex interplay of systems-level brain organization, both during task and resting-state conditions, with relevance for applications in cognitive and clinical neuroscience. The versatility of our methods lets us explore other parts of the central nervous system. In particular, we discovered that spontaneous activity recorded by spinal cord fMRI is highly restless, but can be meaningfully represented by interacting anatomical components. 
Second, the emerging framework of graph signal processing is tailored to neuroimaging by integrating a brain graph (i.e., the structural connectome determined by diffusion-weighted MRI and tractography) and graph signals (i.e., the spatial activity patterns obtained by fMRI). The latter are decomposed onto a graph harmonic basis defined through the eigendecomposition of the graph Laplacian. Spectral filtering operations are then designed to separate brain activity into its structurally aligned and liberal parts, respectively, which allows quantifying of how strongly function is shaped by the underlying structure. The structure-function strength throughout the brain uncovers a behaviorally-relevant spatial gradient from uni- to transmodal regions, which is also informative about task conditions or identifying individuals.
Finally, I will indicate how advances in instrumentation, for instance, ultrafast functional ultrasound, will allow us to surpass current temporal and spatial limitations of fMRI and thus advance analysis and modeling techniques, further contributing to our understanding of the brain at work.