One of the neuroimaging community's challenges is understanding how brain activity is related to cognitive functions. As cognitive functioning requires the brain to allocate resources dynamically to different whole-brain networks, the community has shifted from a more static to different dynamic approaches to analyze functional neuroimaging data and understand cognitive functioning.
In this symposium, we will present the different methods currently being used to analyze functional connectivity dynamics in both fMRI and M/EEG datasets. In doing so, we will present these methods' limitations and potential pitfalls but also showcase how the analysis of functional connectivity dynamics can provide new insights into neurodegenerative diseases like multiple sclerosis.
Dr Qian will present how structure-function decoupling evolves through childhood and underlies cognitive development in youth. Next, Dr. Iraji will introduce the concept of spatial dynamics to incorporate varying spatial configurations into functional connectivity analyses and how this new method translates into clinical applications. One method to capture dynamic brain activity is based on hidden-Markov models (HMMs, Vidaurre et al., Nat Comms 2018), where brain networks with a specific spectral profile – called brain states – are transiently activated. Yet, the main assumption here is that only one brain state is active at any time. To resolve this, Dr. Gohil will present his work on incorporating recurrent neural networks into the analysis of functional connectivity dynamics (Gohil et al. Neuroimage 2022). Finally, Ms Rossi will present her work on HMM-based dynamic brain networks activated during working memory in healthy controls (Rossi et al. Comms Biol 2023) and in people with multiple sclerosis.
Our symposium will highlight both the potential of different functional connectivity approaches to unravel cognitive changes in neuropsychiatric populations, but also highlight the potential pitfalls and challenges that may arise. By the end of the symposium, attendees will (1) be aware of the added value functional connectivity dynamics may bring to characterize dynamic, complex brain functions, (2) have an overview of the different assumptions underlying these methods, and (3) acknowledge the potential of these methods.
• To better understand how functional connectivity is shaped by structural connectivity and yet changes over short and longer time scales
• To have a deeper understanding of potential opportunities, limitations, and pitfalls of current and new analysis techniquess
• To see how FCD can already be applied to improve our knowledge of different neuropsychiatric pathologies
We aim to address our talks to a general HBM audience. This includes both researchers who are developing novel methodologies and would like to have an overview of the most commonly used methods and developments, as well as those more clinically oriented researchers who have to find their way in a jungle of possible algorithms and would like to see practical examples of how the outcomes can be interpreted. Importantly, the methods described in the different talks are applicable to both MEEG and fMRI.
The brain activity supporting high-order cognitive functions, such as attention and working memory, has been extensively described in the spatial configuration (brain regions recruited) via fMRI studies, and in the spectral content and event-related temporal activation in M/EEG studies. However, different neuroimaging techniques and signal/image processing analyses provided scattered descriptions of the cognitive brain processes, which are difficult to integrate.
We present an alternative experimental framework to simultaneously evaluate the three-dimensionalities (time, space, and frequency) of cognitive dynamics. First, we considered MEG data, which provide an optimal temporal resolution (milliseconds) to observe cognitive processes acquired during a working memory paradigm, the visual-verbal nback task. We conducted a dynamic functional network analysis of the data throughout the unsupervised time delay embedded-hidden Markov model (TDE-HMM). This technique identifies a predefined number of spectrally defined patterns that reoccur throughout the recordings, respecting the temporal resolution dictated by the MEG data. In the first stage, we conducted an exploratory analysis on 38 healthy subjects.
This model inferred (working memory) task-specific networks with unique temporal (activation), spectral (phase-coupling connections), and spatial (power spectral density distribution) profiles. First, 200 ms after stimulus onset, a theta frontoparietal network exerts attentional control and encodes the stimulus. Then, an alpha temporo-occipital network rehearses the verbal information, and a broadband frontoparietal network with a P300-like temporal profile leads the retrieval process and motor response.
Following this, we expanded the dataset to a cohort of 70 people with multiple sclerosis (pwMS). We observed that the activation of the early theta prefrontal network significantly decreased in pwMS, which correlated with reduced accuracy in task performance in the MS group, suggesting an impaired encoding and learning process. Secondly, the activation of the M300 frontoparietal network characterized by beta coupling increases in patients treated with benzodiazepine, in line with the well-known benzodiazepine-induced beta enhancement.
Our work on the healthy cohort provides a unified and integrated description of the multidimensional working memory dynamics that can be interpreted within the neuropsychological multi-component model of WM, improving the overall neurophysiological and neuropsychological comprehension of WM functioning. Additionally, the TDE-HMM technique extracted task-relevant functional networks showing disease-specific and treatment-related alterations, revealing potential new markers to assess and track WM impairment in MS. To conclude, combining the MEG task data and a dynamic functional connectivity analysis via TDE-HMM represents the optimal condition to investigate the milliseconds network dynamics underlying any high-order cognitive task in healthy and pathological cases.
, Vrije Universiteit Brussel Brussels, Brussels
Early studies have displayed the success of utilizing resting-state fMRI to quantify brain dynamics, resulting in a growing interest in the evaluation of time-varying functional connectivity patterns. Dynamic analyses are more informative than static analyses and provide richer sets of information, such as the temporal trajectory of functional connectivity patterns obscured by static analyses. Existing work often calculates dynamic functional connectivity using either the same anatomical regions (fixed anatomical patterns) or subject-specific nodes (varying across subjects but spatially fixed over time). However, the dynamic nature of the brain dictates variations in the spatial patterns of brain functional entities over time. In this lecture, I introduce the concept of spatial dynamics and differentiate between different types of dynamic functional connectivity. I provide examples of time-varying spatial patterns and the benefits of incorporating space in functional connectivity analyses, including in the study of clinical populations. I also provide general recommendations for future dynamic functional connectivity research.
, Georgia State University Atlanta, GA
During the development of children and adolescents, the human brain undergoes significant changes in both its structural architecture and functional organization to support increasingly complex cognitive and behavioral capabilities. The brain is functionally organized into large-scale networks along a functional hierarchy extending from unimodal sensory cortex to transmodal association cortex, supporting hierarchical information propagation. This macroscale functional hierarchy is anchored by an anatomical backbone of structural white matter pathways that coordinate synchronized neural activity and cognition. However, only sparse data exist regarding how white matter structural network constraints on brain functional dynamics relate to cognitive improvement during neurodevelopment. In this talk, we will present our work on the examination of the developmental trajectory of brain structure-function relationship in children and adolescents. We used a novel measure – structural-decoupling index to measure the degree of fMRI signal smoothness on structural connectome based on structure-informed graph signal processing (GSP) filtering of functional data. By leveraging on the longitudinal Singapore cohort (GUSTO), we studied how the structure-function decoupling changes longitudinally in preschool children from 4.5 to 6 years old and how the changes support cognitive flexibility maturation. Further, using two independent developmental datasets (NKI and HCP), we identified the developmental trajectories of region-specific structure-function decoupling patterns cross-sectionally and longitudinally in youth from 8 to 22 years old. We will present the differential changes of structure-function decoupling in unimodal and transmodal regions, which led to a spatial decoupling pattern more aligned with the functional hierarchy derived from young adults during the development. We will also present the region- and age-dependent association of structure-function decoupling with cognition during the development. In conclusion, by combining functional dynamics and white matter connectivity, we provide further insights into the neurobiological basis underlying cognitive development.
, National University of Singapore
National University of Singapore
Singapore, Please select an option below
It is now widely accepted that network activity in the brain changes with time. This talk will focus on methodological advancements for identifying dynamic functional networks in neuroimaging data. Time-varying approaches used to model brain networks often assume a mutual exclusivity over time, e.g. sliding window analyses with clustering methods or the Hidden Markov Model (HMM). Whilst a useful constraint, this assumption may compromise the ability of the approach to describe the data effectively. First, I will review existing methods for identifying dynamic brain networks. Following this, I will introduce a new model for neuroimaging data called DyNeMo (Dynamic Network Modes), which is inspired by recent advances in deep learning. This model surpasses existing methods in two ways: the incorporation of a recurrent neural network capable of modelling long-range structure and the ability to describe the data using a time-varying linear mixture of spatially distributed ‘modes’. We demonstrate DyNeMo’s ability to learn mixtures of networks and to model long-range temporal structure on simulated data. Then applying DyNeMo to real magnetoencephalography data, we show DyNeMo infers plausible functional networks with fast dynamics in resting-state data and networks that reflect actions in task data. We show that DyNeMo provides a complementary description to state-based models. Overall, this is a powerful new approach for studying brain network dynamics in both MEG and fMRI datasets.
, University of Oxford Oxford, Oxford