Item

Abstract

Neural activity in the brain is correlated with the blood-oxygen level dependent
(BOLD) contrast which can be measured non-invasively by functional magnetic
resonance imaging (fMRI). Up to date, many fMRI analysis methods are based
on simplifying assumptions about the nature of the BOLD signal. There are two
common assumptions that might lead astray interpretations of experimental re
-
sults: For one, fMRI data has spatial dependencies; each fMRI voxel might be
strongly correlated with some voxels but not with others. These spatial depen
-
dencies are neglected in many analysis methods by assuming statistical indepen
-
dence between voxels. Secondly, the BOLD response to neural activity is not in
-
stantaneous and the exact shape of the hemodynamic response function (HRF)
changes, e.g., across subjects. Most fMRI analyses do not take this variability
into account and assume a canonical HRF.
In this study we employ a recently developed machine learning algorithm to
estimate the spatial correlation structure and the temporal dynamics of the he
-
modynamic response to spontaneous neural activity. We present results from
simultaneous recordings of neural activity and BOLD response in primary vi
-
sual cortex (V1) of the non-human primate. Our results confirm well established
models of the HRF and reveal the spatial correlation structure in V1. The spatial
pattern that correlates best with neural activity can be used to study functional
connectivity. This connectivity measure does not depend on model assumptions
about neural activity or neurovascular coupling mechanisms. In contrast the
connectivity pattern is computed directly from intracranially measured neural
activity and thereby complements existing functional connectivity measures that
are based on fMRI data only.