Optimising Network Modelling Methods for fMri

The activity of functional brain networks is responsible for the emergence of time-varying cognition and behaviour. Accordingly, functional connectivity (FC) in fMRI has been shown to be predictive of behavioural traits and psychiatric and neurological conditions. Many studies estimate FC by calculating the average correlation across an entire fMRI scanning session data (static FC). A major goal of such neuroimaging studies is to develop predictive models to analyze the relationship between whole-brain FC patterns and behavioural traits. However, there is no single widely accepted standard pipeline for analyzing static FC. We proposed six steps for designing FC-based predictive models. This standardized framework for estimating static FC should accelerate both congruence and scientific progress within the neuroscientific community. However, there is also increasing interest in studying the time-varying nature of FC in fMRI (time-varying FC). Typically, methods that measure time-varying FC have several limitations that bias the estimation of time-varying FC, to appear more stable over time than it actually is. We designed a new method, MAGE, that offers a potential explanation and a solution for the homogeneity of time-varying FC seen in existing approaches. This new framework for estimating time-varying FC showed that FC is not static over time and fluctuates on a sub-minute time-scale during resting-state fMRI. This work also investigates the clinical applications of both static and time-varying FC. We found that MAGE estimated time-varying FC can provide a more precise understanding of how brain networks are related to cognitive abilities, both in health and disease. Lastly, we demonstrated thatMAGE can reliably capture the time-varying changes in FC across a range of behavioral and cognitive tasks, overcoming the limitations of most popular existing methods.