Tailored haemodynamic response function increases detection power of fMRI in awake dogs (Canis familiaris)

Abstract

Functional magnetic resonance imaging (fMRI) of awake and unrestrained dogs (Canis familiaris) has been established as a novel opportunity for comparative neuroimaging, promising important insights into the evolutionary roots of human brain function and cognition. However, data processing and analysis pipelines are often derivatives of methodological standards developed for human neuroimaging, which may be problematic due to profound neurophysiological and anatomical differences between humans and dogs. Here, we explore whether dog fMRI studies would benefit from a tailored dog haemodynamic response function (HRF). In two independent experiments, dogs were presented with different visual stimuli. BOLD signal changes in the visual cortex during these experiments were used for (a) the identification and estimation of a tailored dog HRF, and (b) the independent validation of the resulting dog HRF estimate. Time course analyses revealed that the BOLD signal in the primary visual cortex peaks significantly earlier in dogs compared to humans, while being comparable in shape. Deriving a tailored dog HRF significantly improved the model fit in both experiments, compared to the canonical HRF used in human fMRI. Using the dog HRF yielded significantly increased activation during visual stimulation, extending from the occipital lobe, to the caudal parietal cortex, the bilateral temporal cortex, and into bilateral hippocampal and thalamic regions. In sum, our findings provide robust evidence for an earlier onset of the dog HRF in a visual stimulation paradigm, and suggest that using such an HRF will be important to increase fMRI detection power in canine neuroimaging. By providing the parameters of the tailored dog HRF and related code, we encourage and enable other researchers to validate whether our findings generalize to other sensory modalities and experimental paradigms. HighlightsO_LIDog fMRI typically uses human HRF, but underlying neurophysiology might differ C_LIO_LIV1 BOLD signal peaked earlier in dogs than predicted by the human HRF C_LIO_LITailored dog HRF improved model fit when tested with independent data C_LIO_LIWhole-brain comparisons confirmed increased detection power for tailored dog HRF C_LIO_LIDog fMRI will benefit from increased detection power of tailored HRF C_LI