Visualising the brain: advances in neuroimaging analysis

5th Feb 2018

By Tiffany Quinn

The full editorial from 'Brain and Neuroscience Advances' can be downloaded here

Cognition refers to the mental processes underlying our ability to acquire new knowledge and to perceive, understand and think about the world around us. Cognitive neuroscientists therefore seek to better understand the biological processes underlying cognition. For many years neuroimaging techniques such as functional magnetic resonance imaging (fMRI) and electro-encephalography (EEG) have been used to localise cognitive processes in space and time, respectively. There has, however, been some controversy surrounding how much these techniques have contributed to our understanding of the brain’s functional connectivity.

A recent editorial by Williams and Henson (2017), published in the BNA's journal 'Brain and Neuroscience Advances', highlights four new developments in functional neuroimaging analysis that aim to enhance our understanding of the brain. The authors proclaim that the fourth development in particular – data standardisation – will complement the other approaches by: (1) increasing experimental reproducibility and (2) addressing problems with bias and low power results that arise from imaging studies with typically small sample sizes.

The first development discussed is multivariate pattern analysis (MVPA). This statistical technique decodes brain activity across multiple brain regions with high selectivity. At present, it can be used to map the brain and has previously provided evidence of memory suppression and retrieval. The authors are hopeful that in the future, MVPA will be able to explain how changes in different brain regions relate to each other, which may enable closer mapping to artificial networks of cognitive processes.

Evidence now suggests that the functional connectivity of the brain varies over time. Time-varying approaches have therefore been developed to quantify the properties of these dynamic networks during cognitive tasks. This means that they can be used to characterise different states of brain connectivity, which can be used to develop cognitive theories and provide insight into various neurological or psychiatric conditions.

Neurobiological modelling - based on the neuroimaging data - has been developed to simulate the neural mechanisms underpinning brain development and disease. For example, the deduction of parameters such as neurotransmitter concentration within particular populations can be deduced with neural mass models of fMRI data. In this way, it is hoped that the information deduced from neuroimaging data will bridge the gap between other methods of neuroscientific investigations. Additionally, simple models of resting-state networks are useful tools for defining certain properties. Despite their promising potential, these techniques are in their infancy and are thus limited in what they can capture at present.

The last development discussed in this review is the standardisation of big data before collection that, the authors believe, will be very beneficial with regards to increasing reproducibility and therefore reliability. It is hoped that a continued encouragement of standard practice will not prevent the development of innovative methods.

Neuroimaging data is vast and complex; the advances in the analytical methods discussed in this review demonstrate that progress is underway in obtaining more detailed information about the structure and function of the brain and its neural processes. The implication of this progress is a better appreciation of the differences between the workings of the brain in health and disease and one would hope that this serves to enhance medical research and treatment strategies. 

The full editorial from 'Brain and Neuroscience Advances' can be downloaded here