What aspect of the fMRI BOLD signal best reflects the underlying electrophysiology in human somatosensory cortex?

https://doi.org/10.1016/S1388-2457(03)00080-4Get rights and content

Abstract

The interpretation of task-induced functional imaging of the brain is critically dependent on understanding the relationship between observed haemodynamic responses and the underlying neural changes. However, the precise nature of this neurovascular coupling relationship remains unknown. In particular, it is unclear which measure of functional magnetic resonance imaging blood oxygen level dependent (fMRI BOLD) activity is the best correlate of neural activity. We measured the somatosensory evoked potential (SEP) amplitude at the scalp, and fMRI BOLD signal to increases in intensity of contralateral median nerve electrical stimulation in healthy non-anaesthetised subjects. We compared correlation analyses between SEP amplitude and both peak voxel fMRI BOLD percentage signal change and mean voxel fMRI BOLD percentage signal change across a somatosensory cluster, and we also performed a voxel-by-voxel correlation between fMRI BOLD activity and SEP amplitude. We found that fMRI BOLD changes in primary somatosensory cortex correlate significantly with SEP amplitudes, suggesting a linear neurovascular coupling relationship under the conditions investigated. We also found that mean changes across a cluster correlate less well with SEP amplitude than peak voxel levels. This suggests that the area of haemodynamic activity correlating with SEP amplitude is smaller than the entire cluster observed.

Introduction

The interpretation of task-induced functional imaging of the brain is critically dependent on understanding the relationship between observed haemodynamic responses and the underlying neuronal changes. We have established a linear relationship between peak-voxel functional magnetic resonance imaging blood oxygen level dependent (fMRI BOLD) signals and somatosensory evoked potential (SEP) amplitude in human somatosensory cortex (Arthurs et al., 2000). This is consistent with the suggestion that fMRI BOLD responses are able to reflect underlying synaptic changes, as short-latency SEPs are attributed mainly to synchronized extracellular currents from summated post-synaptic potentials on pyramidal cells (Eccles, 1951, Lopes da Silva and Storm van Leeuwan, 1978, Allison et al., 1991, Lopes da Silva, 1991). They are highly reproducible and are generally resistant to changes in attention and accommodation (Mauguiere, 1995).

However, the precise nature of this neurovascular coupling relationship remains unknown. In particular, it is unclear which measure of fMRI BOLD activity is the best correlate of neural activity. In order to examine this issue, we measured the SEP amplitude at the scalp, and fMRI BOLD signal to increases in intensity of contralateral median nerve electrical stimulation in healthy non-anaesthetised subjects. Here we present a comparison of the correlation between SEP amplitude and both peak voxel fMRI BOLD percentage signal change and mean voxel fMRI BOLD percentage signal change across a somatosensory cluster. We also performed a voxel-by-voxel correlation between fMRI activity and SEP amplitude. Some of this data has been presented previously (Arthurs et al., 2000, Arthurs et al., 2001).

Section snippets

Stimulation

Volunteers (n=6; 4 males; mean age 24.33 years, range 22–29) were healthy adults recruited from local University members. All studies were performed under Local Ethics Committee Approval guidelines, with informed consent obtained from all volunteers. Stimuli were 0.2 ms square-wave electrical pulses delivered to the median nerve at the wrist, at 25–175% of predetermined motor threshold. Stimulation did not exceed 30 mA or individual pain thresholds. fMRI responses were measured at 100 Hz

Results

N20–P25 and P25–N30 amplitudes of the cortical SEP significantly increased with stimulus intensity in all subjects examined (P<0.01, P<0.05, respectively; group data P<0.001). Largest amplitudes were reached at 125% of motor threshold, which showed marked inter-individual variability. SEP peak latencies did not vary significantly (P<0.4; not shown).

Using voxel-by-voxel t-statistics across all intensities, the most significant clusters of activation in fMRI BOLD were observed in contralateral

Discussion

We used stimulus intensity to change both SEP amplitudes and fMRI BOLD signal changes in somatosensory cortex in parallel. Our key finding is that SEP amplitudes correlate linearly with fMRI BOLD changes in somatosensory cortex across the intensities used, suggesting that the fMRI responses mirror the underlying electrophysiological measures, i.e. a close neurovascular coupling relationship (under the conditions investigated). This is important in the interpretation of human fMRI BOLD signals,

Acknowledgements

We would like to thank the whole team of the Wolfson Brain Imaging Centre, including T. Donovan, V. Lupson, T.A. Carpenter and its chairman J.D. Pickard. Technology Foresight, UK, supported this project with Oxford Instruments, UK. Merck Sharp and Dohme support O.J.A. on a MB/PhD program.

References (26)

  • O.J. Arthurs et al.

    Comparison of mean and peak voxel fMRI signal changes against SEPs for quantitative neurovascular coupling

    Clin Neurophysiol

    (2001)
  • G. Brinker et al.

    Simultaneous recording of evoked potentials and T2*-weighted MR images during somatosensory stimulation of rat

    Magn Reson Med

    (1999)
  • R.B. Buxton et al.

    Dynamics of blood flow and oxygenation changes during brain activation: the balloon model

    Magn Reson Med

    (1998)
  • Cited by (86)

    • Fully integrated wide dynamic range optical receiver for near infrared spectroscopy

      2019, Microelectronics Journal
      Citation Excerpt :

      NIRS is proved to be a non-invasive, safe, and relatively cheap technique which monitors cerebral hemodynamic changes associated with different conditions. Furthermore, several researches has investigated the ability of NIRS to predict cardiovascular risk [2], diagnose skin lesion [3], measure lung efficiency [4], and construct a brain functions map [5]. However, there are different challenges that faces the design of a solid NIRS optical receiver.

    • High-resolution fMRI investigations of the fingertip somatotopy and variability in BA3b and BA1 of the primary somatosensory cortex

      2016, Neuroscience
      Citation Excerpt :

      Although the somatotopy in BA3b was similar for the peak value and the center of mass approach, we did not find a somatotopy in BA1 if the peak value approach was applied. This contrasts with previous investigations which found that peak value methods are the best approaches for those investigations (Arthurs and Boniface, 2003; Zhang et al., 2010; Schweisfurth et al., 2011), but presumably took into account only BA3b. The difference between BA3b and BA1 is reasonable assuming an orientation column-like structure in BA1 as found in monkey investigations (Friedman et al., 2004), which increases the demand on the fMRI analysis due to the larger size of the activated areas and the small variation of the amplitudes designating the representation of a fingertip.

    • Functional Connectivity Is Associated with Altered Brain Chemistry in Women with Endometriosis-Associated Chronic Pelvic Pain

      2016, Journal of Pain
      Citation Excerpt :

      In this analysis, CSF-adjusted levels of excitatory neurotransmitters, BPI pain intensity, STPIDA-A, and STPIDA-D were used as the independent variables, and functional connectivity data (seed correlation and variance maps) were used as the dependent variables. Functional connectivity values (z statistic) were extracted from the peak voxel, rather than all voxels of the ROI, because peak voxels have previously been demonstrated to have stronger correlation with synaptic activity.4 All brain maps were thresholded using the standard FSL toolbox cluster correction for multiple comparisons.

    View all citing articles on Scopus
    View full text