Castaldo, Francesca; (2024) Modelling large-scale cortico-subcortical networks: Whole-brain Models of Transient Oscillatory Patterns. Doctoral thesis (Ph.D), UCL (University College London).

Text Castaldo_10191094_Thesis_final.pdf Access restricted to UCL open access staff until 1 May 2025. Download (23MB)

Abstract

Since the dawn of the 20th century, neuroscientists have unearthed rhythmic oscillations in the mammalian brain's electrical activity, captured through techniques like electroencephalography (EEG) and its magneto-counterpart (MEG). These oscillatory patterns have been linked to various physiological states and pathological conditions. Yet, a comprehensive understanding of the underlying mechanisms orchestrating these brain rhythms remains an enigmatic frontier, hindering progress in devising targeted therapeutic interventions. Initiating our inquiry, we re-examine the hypothesis that transient neural oscillations emerge from a balance of synchronisation among disparate cerebral regions. These are thought to operate at reduced collective frequencies due to inherent signal propagation delays. Extending previous phase-oscillator models, the amplitude dynamics is included to approximate input-driven gamma frequency oscillations in local fields. In line with analytic predictions, we identify a critical regime where spatially and spectrally resolved metastable oscillatory modes emerge spontaneously from weak synchronisation of distinct subsystems, approximating the MEG power spectra from awake healthy humans. Existing whole-brain models are generally tailored to modelling a particular data modality (e.g., fMRI or MEG/EEG). Despite the differing aspects of neural activity, each modality captures, we propose that they originate from shared network dynamics. To validate this, we deploy two large-scale models — Stuart-Landau and Wilson-Cowan — to jointly predict diverse neural attributes across fMRI and MEG modalities. Our analyses highlight emergent features like metastable oscillatory modes and functional connectivity dynamics, albeit with nuanced differences. Notably, dismissing conduction delays or equilibrium shifts between excitation and inhibition adversely impacts model accuracy. Despite limitations in cross-modality correlation, our models demonstrate the emergence of functional connectivity patterns extending beyond anatomical constraints and highlight the role of mesoscale heterogeneities in neural circuitry. This unified approach sheds light on the universal principles underpinning different aspects of neural activity and offers a foundation for future multi-modal modelling efforts. In the last part of our work, we exploit computational simulations to verify the sensitivity of large-scale models to external perturbations. We delve into the network dynamics under diverse stimulation protocols, ranging from unilateral step function to sinusoidal inputs. Various metrics, including but not limited to power spectral density and functional connectivity distance, are employed to dissect the modulated behaviour of specific nodes within these neural circuits. We find that constant stimuli amplify global network coherence, while sinusoidal stimuli generate more compartmentalised neural activities. Furthermore, we unveil a robust statistical linkage between a node's structural degree and its average activation level during external stimulation. Regardless of the nature of the stimulus, nodes exhibiting higher structural connectivity consistently manifest greater functional connectivity modulation. Similarly, we introduced a complementary framework that considers bilateral sinusoidal stimulation patterns associated with clinical improvement after deep brain stimulation in the subthalamic nucleus (STN-DBS) for three different pathologies. Our main finding highlights a consistent difference in functional connectivity before and during perturbation across each clinical category. This observation is further substantiated by a hierarchical clustering analysis that recorded a strength of approximately 92%. This underlines a pronounced differentiation in functional connectivity patterns among distinct clinical STN-DBS sweet-tracts categories, emphasising the efficacy of sweet-tracts-dependent perturbation in modulating connectivity. This multi-faceted exploration elucidates the nuances dictating neural network responses to external modulations, thereby directly impacting theoretical neuroscience and prospective clinical applications.

Download activity - last month

Export as CSVJSONXML

Download activity - last 12 months

Export as CSVJSONXML

Downloads by country - last 12 months

Export as XMLJSONCSV

Archive Staff Only

View Item