Processing of brain time series to unravel changes in dynamic functional connectivity in disease

点击翻译按钮获取中文摘要

This paper aims to describe the applicant's research approach on the subject of implementing new analysis and computational methods on acquired brain signals. Research in this area has started to emphasize the importance of getting a connectivity-based understanding of the relationship between fast time series events in the brain. In order to do this, computational models are being built as an approximation of brain functionality. The traditional, "stationary" approach to understanding brain dynamics is to consider the structural connectivity of the brain as the main neural activity mapping means in trying to identify the networks responsible for certain cognitive activities, as proposed in the 1980s. More modern approaches recognise the non-linear and evasive nature of neural paths while also taking into consideration the anatomical structure (more recently called the connectome). This is as they recognize that brain activity varies in time and space dependent on the type of activity the brain is involved in and on the mental state (Fornito et al., 2016) [1]. Since this historic milestone in neuroscience, research in the area of aim to create a correct approximation of the spatiotemporal nature of brain signals and build a coherent functional and effective connectivity model. A few main challenges could be identified in this direction:1. Moving away from brain modularity and instead identifying spontaneous and task-related connectivity general patterns. The model must also account for individual variability. This can be defined as the functional connectivity problem. To tackle this challenge, models are built to describe or predict neural pattern activation as response to an applied simulated excitation input, without emphasizing time dependency between different regions of the brain [2], [3]. 2. Identifying the causal relationship between statistically dependent signals - i.e. defining the effective connectivity (or dynamic functional connectivity), a complex mathematical and computational task. To overcome this challenge, models are built which allow a trial-and-error approach to identifying the cause of the studied brain activity [2], [3]. Unlike the previously mentioned type of models, this line of action emphasizes the causal relationship between activatedneural regions