False activation in the brain ventricles related to task-correlated breathing in fMRI speech and motor paradigms

Abstract

We demonstrate, and show how to eliminate, a task-correlated breathing activation artefact created while performing exercise tasks during a gap in functional magnetic resonance imaging (fMRI). Two studies are presented. The first was intended to isolate a reliable fMRI paradigm for intense handgrip contractions. A gapped acquisition was used to reduce motion artefact, where the contraction was performed during a gap, and image acquisition was between contractions. The second study involved naming regular words (REGs) and nonwords (NWs), where a gap is required for the analysis of participants' overt speech. For study 1, brain ventricle activation was present when breathing responses were task-correlated, and was only eliminated by removing the gap from the sequence. For study 2, NWs were associated with activation artefact in the ventricles, and slower reaction time (RT), reflecting a strategy whereby breathing falls in synchrony with image acquisition. REGs showed the expected RT distribution and frequency effect (reflecting lexical access), with no ventricle activation, and consequently no synchrony with image acquisition. The gapped paradigm increased the likelihood of breathing correlated T2* signal changes in brain ventricles. FMRI researchers should examine the brain ventricles for activation artefact as they are likely associated with false activations in other brain regions.