Elsevier

NeuroImage

Volume 153, June 2017, Pages 369-381
NeuroImage

Review
The relationship between BOLD fMRI response and the underlying white matter as measured by fractional anisotropy (FA): A systematic review

https://doi.org/10.1016/j.neuroimage.2016.12.075Get rights and content

Highlights

  • The relationship between BOLD and FA appears to be task and region dependent.

  • There is a mostly positive correlation between FA and BOLD in primary sensory and motor regions.

  • Both positive and negative correlations are observed between FA and BOLD in cognitive tasks.

  • It has to be distinguished whether the reported relationship is stemming from healthy or patient groups.

  • Behavioural and clinical data are essential for interpreting the relationships between FA and BOLD.

Abstract

Despite the relationship between brain structure and function being of fundamental interest in cognitive neuroscience, the relationship between the brain's white matter, measured using fractional anisotropy (FA), and the functional magnetic resonance imaging (fMRI) blood oxygen level dependent (BOLD) response is poorly understood. A systematic review of literature investigating the association between FA and fMRI BOLD response was conducted following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. The PubMed and Web of Knowledge databases were searched up until 22.04.2016 using a predetermined set of search criteria. The search identified 363 papers, 28 of which met the specified inclusion criteria. Positive relationships were mainly observed in studies investigating the primary sensory and motor systems and in resting state data. Both positive and negative relationships were seen in studies using cognitive tasks. This systematic review suggests that there is a relationship between FA and the fMRI BOLD response and that the relationship is task and region dependent. Behavioural and/or clinical variables were shown to be essential in interpreting the relationships between imaging measures. The results highlight the heterogeneity in the methods used across papers in terms of fMRI task, population investigated and data analysis techniques. Further investigation and replication of current findings are required before definitive conclusions can be drawn.

Introduction

The relationship between structure and function in the human brain is of fundamental interest in cognitive neuroscience. Advances in human brain magnetic resonance imaging (MRI) have lead to a variety of tools being at the researcher's disposal for investigating such relationships. Particular progress has been made in the combination of different imaging methods (Shah et al., 2013, Zhu et al., 2014). The wealth of data obtained using such multimodal imaging approaches allows multiple aspects of human brain structure and function to be investigated in the same individuals. This affords the exploration of relationships between these measures in both healthy brains and brain pathologies.

At this point it is necessary to define the terms used for the purposes of this systematic review. We investigate structure and function on a macro scale (Honey et al., 2010): voxels of several cubic millimetres are used in the brain imaging methods under investigation. Additionally, by “structure” we mean the anatomical organization of the brain, specifically white matter. By “function” we mean the haemodynamic of the brain as captured by functional MRI (fMRI) after enhanced neural activity (Logothetis, 2008). We further extend this to the behavioural expression of function in terms of psychological, clinical or performance measures.

Structural connections between brain regions are important for understanding functional interactions in these areas (Friston and Frackowiak, 1997). Much of the research to date investigates the relationship between structural connectivity and functional connectivity (see Damoiseaux and Greicius (2009), and Honey et al. (2010), for reviews). However, it is not only the presence, or strength, of connections between brain regions that is important but also the microstructural characteristics of the tissue forming these connections (Camara et al., 2010). White matter structures allow the efficient propagation of neural signals between spatially distinct cortical regions (Waxman et al., 1995); thus they can influence the neural dynamics in terms of how efficiently information is passed along these tracts. Based on this, it would be also reasonable to assume that the microstructural properties of these white matter connections between brain regions are related to the function of those regions. This is supported to some extent in the literature where deterioration in measures of white matter integrity are observed in some diseases and disorders such as chronic pain (Moayedi et al., 2012, Ellingson et al., 2013), Alzheimer's (Amlien and Fjell, 2014) and schizophrenia (Marenco et al., 2012). A reduction in white matter integrity has also been associated with a decline in executive function in normal aging (Sasson et al., 2013), motor function (Sullivan et al., 2010) and sensory-motor gating (Ota et al., 2013). The impact of these white matter changes on observable clinical and behavioural outcomes suggests that brain function is also related to the white matter.

White matter and the functional brain response can be measured using MRI techniques. A powerful tool for investigating white matter (WM) is Diffusion Tensor Imaging (DTI, Basser et al., 1994). The commonly used scalar metric of DTI is fractional anisotropy (FA). FA characterizes the directionality of constrained water diffusion in the brain tissue (Song et al., 2003). However, it should be considered that terms such as “white matter integrity” or “microstructural integrity” are often used relatively vague in studies applying DTI, e.g. they are sometimes confused with FA. Moreover, these terms are especially misleading when the reported findings relate to healthy volunteers (for a more detailed discussion see Jones et al. (2013)). The most common measure of brain function in fMRI is the blood oxygen level dependent (BOLD) contrast (Ogawa and Lee, 1990, Ogawa et al., 1990). The BOLD contrast primarily relies on magnetic properties of haemoglobin that differ depending on whether it is bound to oxygen, thus it measures the metabolic demands of active neurons rather than the neural activity per se. Percentage signal change, or similar quantification of the BOLD response, is used to assess functional responses to stimuli or to assess the level of brain activation during rest.

Employing a multimodal imaging approach to investigating the relationship between measures of brain structure and function is a logical progression as the techniques mature. As such, a growing number of studies using measures of both white matter (DTI, FA) and measures of grey matter function (BOLD fMRI) have been performed. A number of approaches exist for integrating DTI and fMRI data. A review by Zhu et al. (2014) described three approaches: 1. fMRI assists DTI, for example fMRI guided fibre tracking or fMRI based validation of DTI results; 2. DTI assists fMRI, for example functional connectivity analysis based on DTI data; 3. Joint DTI/fMRI fusion, for example deriving results from each modality separately and combining them using statistical analysis (e.g. t-test, correlation) or initially integrating both sets of data and then analysing using a hybrid model. It is the third approach that we will focus on. Specifically, this systematic review will cover papers where this approach is used to investigate a relationship between the BOLD response in cortical regions and FA in the underlying WM.

Despite the relationship between structure and function being of fundamental interest in many areas of research and clinical neuroscience, the relationship between FA and the fMRI BOLD has not been comprehensively investigated. Given that decreased FA would suggest increased diffusion in the direction perpendicular to the main fibre orientation, in other words, more isotropic diffusion indicating less organisation (Davis and Moayedi, 2013), FA values might reflect how efficiently information is transmitted and thus be related to the functional response to a stimulus (Ota et al., 2013). However, the findings so far are equivocal as to whether FA and fMRI BOLD response are statistically related and if so, whether they are positively or negatively related. The principle aim of this systematic review is to establish whether there is a relationship between cortical fMRI BOLD response and FA in the underlying white matter, and whether this is a positive or negative relationship. A supplementary aim is to investigate whether these imaging measures are related to behavioural or psychological measures. To this end we review published papers that measured and quantified both FA and fMRI BOLD response and made a statistical assessment of the relationship between the two measures.

Section snippets

Methods

The systematic review was conducted in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines (Moher et al., 2009) using an advanced document protocol (see S1). The advanced protocol document was not registered since the requirement for investigating health/clinical outcomes (e.g. for registration with PROSPERO, http://www.crd.york.ac.uk/prospero/) was not fulfilled. The advanced document protocol was updated once to include an additional

Results

The search strategy yielded a total of 416 papers after removing duplicates, 363 of which were removed at the screening stage based on the inclusion and exclusion criteria described in the methods section: 240 did not use DTI, 30 of the papers using DTI did not report FA, 3 papers did not use fMRI, 7 used non human subjects, 42 were review articles, 7 were case studies, 2 were book chapters, 16 only existed as abstracts or conference proceedings, 9 were not written in English, 5 were technical

Discussion

We performed a systematic review of literature investigating the relationships between FA and BOLD response in the human brain. In relation to our principle aim, to establish whether there is a relationship between cortical fMRI BOLD response and FA in underlying white matter, we found positive, negative and non-significant relationships between FA and BOLD measures in the reviewed literature. Positive relationships were mainly observed in studies investigating the primary sensory and motor

Funding

There are no funding sources or conflicts of interest to declare.

References (52)

  • P.J. Olesen et al.

    Combined analysis of DTI and fMRI data reveals a joint maturation of white and grey matter in a fronto-parietal network

    Brain Res. Cogn. Brain Res.

    (2003)
  • M. Ota et al.

    Multimodal image analysis of sensorimotor gating in healthy women

    Brain Res.

    (2013)
  • N.J. Shah et al.

    Advances in multimodal neuroimaging: hybrid MR-PET and MR-PET-EEG at 3 T and 9.4 T

    J. Magn. Reson.

    (2013)
  • K.D. Singh et al.

    Transient and linearly graded deactivation of the human default-mode network by a visual detection task

    Neuroimage

    (2008)
  • S.K. Song et al.

    Diffusion tensor imaging detects and differentiates axon and myelin degeneration in mouse optic nerve after retinal ischemia

    Neuroimage

    (2003)
  • E.V. Sullivan et al.

    Fiber tracking functionally distinct components of the internal capsule

    Neuropsychologia

    (2010)
  • S.J. Teipel et al.

    White matter microstructure underlying default mode network connectivity in the human brain

    Neuroimage

    (2010)
  • A.T. Toosy et al.

    Characterizing function-structure relationships in the human visual system with functional MRI and diffusion tensor imaging

    Neuroimage

    (2004)
  • G. Wagner et al.

    Structural and functional dysconnectivity of the fronto-thalamic system in schizophrenia: a DCM-DTI study

    Cortex

    (2015)
  • T. Warbrick et al.

    Using structural and functional brain imaging to investigate responses to acute thermal pain

    J. Pain

    (2016)
  • Y. Zhou et al.

    Altered resting-state functional connectivity and anatomical connectivity of hippocampus in schizophrenia

    Schizophr. Res.

    (2008)
  • D. Zhu et al.

    Fusing DTI and fMRI data: a survey of methods and applications

    Neuroimage

    (2014)
  • Z. Zhu et al.

    Age-related increases in right frontal activation during task switching are mediated by reaction time and white matter microstructure

    Neuroscience

    (2014)
  • S.V. Astafiev et al.

    Abnormal white matter blood-oxygen-level-dependent signals in chronic mild traumatic brain injury

    J. Neurotrauma

    (2015)
  • A.A. Baird et al.

    Functional connectivity: integrating behavioral, diffusion tensor imaging, and functional magnetic resonance imaging data sets

    J. Cogn. Neurosci.

    (2005)
  • V. Bonnelle et al.

    Salience network integrity predicts default mode network function after traumatic brain injury

    Proc. Natl. Acad. Sci. USA

    (2012)
  • Cited by (0)

    1

    Contributed equally.

    View full text